[go: up one dir, main page]

WO2020162425A1 - Analysis device, analysis method, and program - Google Patents

Analysis device, analysis method, and program Download PDF

Info

Publication number
WO2020162425A1
WO2020162425A1 PCT/JP2020/004042 JP2020004042W WO2020162425A1 WO 2020162425 A1 WO2020162425 A1 WO 2020162425A1 JP 2020004042 W JP2020004042 W JP 2020004042W WO 2020162425 A1 WO2020162425 A1 WO 2020162425A1
Authority
WO
WIPO (PCT)
Prior art keywords
discriminator
analysis
production facility
data
detection result
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/JP2020/004042
Other languages
French (fr)
Japanese (ja)
Inventor
康晴 大西
靖行 福田
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
NEC Corp
Original Assignee
NEC Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by NEC Corp filed Critical NEC Corp
Priority to KR1020217024139A priority Critical patent/KR20210107844A/en
Priority to PH1/2021/551623A priority patent/PH12021551623A1/en
Priority to CN202080012631.9A priority patent/CN113383216A/en
Priority to JP2020571198A priority patent/JP7188463B2/en
Publication of WO2020162425A1 publication Critical patent/WO2020162425A1/en
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01H—MEASUREMENT OF MECHANICAL VIBRATIONS OR ULTRASONIC, SONIC OR INFRASONIC WAVES
    • G01H1/00—Measuring characteristics of vibrations in solids by using direct conduction to the detector
    • G01H1/12—Measuring characteristics of vibrations in solids by using direct conduction to the detector of longitudinal or not specified vibrations
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01H—MEASUREMENT OF MECHANICAL VIBRATIONS OR ULTRASONIC, SONIC OR INFRASONIC WAVES
    • G01H17/00—Measuring mechanical vibrations or ultrasonic, sonic or infrasonic waves, not provided for in the preceding groups
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M99/00—Subject matter not provided for in other groups of this subclass
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion

Definitions

  • the present invention relates to an analysis device, an analysis method, and a program, and more particularly to an analysis device, an analysis method, and a program for analyzing data of a sensor that monitors the state of equipment.
  • vibration and acoustic sensors for manufacturing quality control of production materials using mechanical equipment. For example, by acquiring vibration data generated by a processing machine during processing of production materials and avoiding production loss by stopping the processing of production materials when abnormal vibrations are captured, the operating status of production equipment can be used as vibration. It is attracting attention as a technology that improves the production efficiency of the manufacturing industry by monitoring and monitoring it, improving the efficiency of maintenance, and finding the optimal operating conditions for extending the life of the equipment.
  • Patent Document 1 discloses a method in which a sensor is attached to a facility to be monitored and the facility is monitored based on time series data measured by the sensor.
  • the present invention has been made in view of the above circumstances, and an object of the present invention is to provide a technique for efficiently and accurately monitoring the state of production equipment.
  • the first aspect relates to an analysis device.
  • the first analysis device is Image processing means for imaging the detection result of the vibration sensor provided in the production facility, Generating means for generating a discriminator by subjecting the imaged data to machine learning processing; And an analysis unit that performs a state analysis process of the production facility using the discriminator.
  • the second analysis device Determination means for performing abnormality determination processing of the production facility using the first discriminator based on the detection result of the vibration sensor provided in the production facility; Image processing means for imaging the detection result, which cannot be discriminated as normal or abnormal by the first discriminator; Generating means for generating a second discriminator using the imaged data as a target of machine learning processing; Analysis means for performing a state analysis process of the production facility using the second discriminator.
  • the second aspect relates to at least one computer implemented analysis method.
  • the first analysis method according to the second aspect is The analysis device Visualize the detection result of the vibration sensor installed in the production facility, Generate a discriminator with the imaged data as the target of machine learning processing, Performing state analysis processing of the production facility using the discriminator.
  • the second analysis method according to the second aspect is The analysis device Based on the detection result of the vibration sensor provided in the production equipment, the abnormality determination processing of the production equipment is performed using the first discriminator, The detection result, which could not be discriminated as normal or abnormal by the first discriminator, is imaged, A second discriminator is generated by using the imaged data as a target of machine learning processing, Performing state analysis processing of the production facility using the second discriminator.
  • the present invention may be a program that causes at least one computer to execute the method of the second aspect, or a computer-readable recording medium that records such a program. May be.
  • the recording medium includes a non-transitory tangible medium.
  • the computer program includes a computer program code that, when executed by a computer, causes the computer to perform the analysis method on an analysis device.
  • the various constituent elements of the present invention do not necessarily have to be independently present, and a plurality of constituent elements are formed as one member, and one constituent element is formed by a plurality of members. May be present, a certain component may be a part of another component, a part of a certain component may overlap a part of another component, and the like.
  • the order of description does not limit the order in which the plurality of procedures are executed. Therefore, when carrying out the method and computer program of the present invention, the order of the plurality of procedures can be changed within a range that does not hinder the contents.
  • the plurality of procedures of the method and computer program of the present invention are not limited to being executed at different timings. For this reason, another procedure may occur during the execution of a certain procedure, the execution timing of a certain procedure and the execution timing of another procedure may partially or entirely overlap, and the like.
  • FIG. 4 is a diagram showing time series data of measurement data before imaging. It is a figure which shows the image data imaged by the image processing part. It is a flowchart which shows an example of operation
  • 17 is a flowchart illustrating an example of a discriminator update processing procedure using the data extracted in step S406 of FIG. 16. 17 is a flowchart showing an example of a processing procedure when it is determined to be normal in step S401 of the state analysis processing of FIG. It is a flow chart which shows an example of the procedure of defective model construction processing of an analysis device.
  • FIG. 3 is a flowchart for explaining the analysis device of the first embodiment.
  • FIG. 6 is a flowchart for explaining the analysis device of the second embodiment.
  • FIG. 1 is a diagram conceptually showing a system configuration of an equipment monitoring system 1 using an analysis device according to an embodiment of the present invention.
  • the equipment monitored by the equipment monitoring system 1 is the production equipment 10, and in the present embodiment, a belt conveyor will be described as an example.
  • a plurality of sensors 12 for monitoring the belt conveyor are installed at a plurality of locations along the moving direction of the belt conveyor.
  • Each sensor 12 is, for example, a vibration sensor.
  • a vibration sensor that detects vibration in one direction
  • a plurality of vibration sensors may be installed at one location in order to detect vibration in multiple directions.
  • the measurement data output from the vibration sensor is time series data indicating a vibration waveform.
  • the vibration sensor measures the vibration generated in the production facility 10 to be monitored.
  • the vibration sensor may be a uniaxial acceleration sensor that measures acceleration in the uniaxial direction, a triaxial acceleration sensor that measures acceleration in the triaxial directions, or any other type.
  • the plurality of vibration sensors may be the same kind of vibration sensor, or a plurality of kinds of vibration sensors may be mixed.
  • the analysis device 100 is connected to the GW (GateWay) 5 via the network 3 and receives detection results from the plurality of sensors 12 provided in the production facility 10.
  • the analysis device 100 is connected to the storage device 20.
  • the storage device 20 stores vibration data analyzed by the analysis device 100.
  • the storage device 20 may be a device separate from the analysis device 100, a device included in the analysis device 100, or a combination thereof.
  • FIG. 2 is a diagram showing an example of a data structure of the vibration data 22 and the facility information 24 stored in the storage device 20 of this embodiment.
  • the vibration data 22 is associated with time information and measurement data for each sensor ID that identifies the vibration sensor.
  • the facility information 24 is associated with the sensor ID of at least one vibration sensor for each facility ID that identifies the facility.
  • FIG. 3 is a block diagram illustrating the hardware configuration of each device of this embodiment.
  • Each device has a processor 50, a memory 52, an input/output interface (I/F) 54, a peripheral circuit 56, and a bus 58.
  • the peripheral circuit 56 includes various modules.
  • the processing device may not have the peripheral circuit 56.
  • the bus 58 is a data transmission path for the processor 50, the memory 52, the peripheral circuit 56, and the input/output interface 54 to mutually transmit data.
  • the processor 50 is an arithmetic processing device such as a CPU (Central Processing Unit) and a GPU (Graphics Processing Unit).
  • the memory 52 is a memory such as a RAM (Random Access Memory) or a ROM (Read Only Memory).
  • the input/output interface 54 includes an interface for acquiring information from an input device, an external device, an external server, a sensor, etc., an interface for outputting information to an output device, an external device, an external server, etc.
  • the input device is, for example, a keyboard, a mouse, a microphone, or the like.
  • the output device is, for example, a display, a speaker, a printer, a mailer, or the like.
  • the processor 50 can issue a command to each module and perform a calculation based on the calculation result.
  • Each component of the analysis device 100 of this embodiment of FIG. 4 described later is realized by an arbitrary combination of the hardware and software of the computer shown in FIG. It will be understood by those skilled in the art that there are various modified examples of the realizing method and the apparatus.
  • the functional block diagram showing the analysis device of each embodiment described below shows blocks of logical functional units, not of hardware units.
  • each function of each unit of the analysis device 100 in FIG. 4 can be realized.
  • the computer program of the present embodiment causes a computer (processor 50 in FIG. 3) for realizing the analysis device 100 to display a procedure for imaging the detection result of the vibration sensor 12 provided in the production facility 10 and the imaged data. It is described that a procedure for generating a discriminator as a target of machine learning processing and a procedure for performing an abnormality determination processing of the production facility 10 using the discriminator are executed.
  • the computer program of this embodiment may be recorded in a computer-readable recording medium.
  • the recording medium is not particularly limited, and various forms are conceivable.
  • the program may be loaded from the recording medium into the memory 52 (FIG. 3) of the computer, or may be downloaded to the computer through the network and loaded into the memory 52.
  • a recording medium for recording a computer program includes a non-transitory tangible computer-usable medium, and a computer-readable program code is embedded in the medium.
  • the computer program When the computer program is executed on the computer, it causes the computer to execute the analysis method of the present embodiment that realizes the analysis device 100.
  • FIG. 4 is a functional block diagram showing a logical configuration of the analysis device 100 of this embodiment.
  • the analysis device 100 includes an image processing unit 102, a generation unit 104, and an analysis unit 106.
  • the image processing unit 102 visualizes the detection result of the vibration sensor provided in the production facility 10.
  • the generation unit 104 generates the discriminator 110 by using the imaged data as the target of machine learning processing.
  • the analysis unit 106 uses the discriminator 110 to perform a state analysis process of the production facility 10.
  • the measurement data of the vibration sensor is time-series data including a plurality of vibration waveforms indicated by a plurality of parameters, the characteristics cannot be captured even by machine learning as it is, and an appropriate learning model cannot be generated. Therefore, in the present embodiment, the measurement data is imaged by the image processing unit 102.
  • FIG. 5 is a diagram showing time-series data of measurement data before imaging.
  • FIG. 6 is a diagram showing image data imaged by the image processing unit 102.
  • the image processing unit 102 acquires a detection result of the sensor 12 of the production facility 10 (hereinafter, also referred to as vibration data). Then, the image processing unit 102 performs FFT processing on the acquired vibration data, frequency-divides the acquired vibration data, and images the acquired vibration spectrum data to obtain image data. By these processes, the data capacity can be compressed, and the machine learning process and the discrimination process can be speeded up.
  • acquisition means that the device itself acquires data or information stored in another device or a storage medium (active acquisition), and is output to the device itself from another device. At least one of inputting data or information (passive acquisition) is included. Examples of active acquisition include requesting or inquiring another device to receive the reply, and accessing and reading the other device or a storage medium. An example of passive acquisition is receiving information to be distributed (or transmitted, push notification, etc.). Further, “acquisition” may mean selecting and acquiring from received data or information, or selecting and receiving distributed data or information.
  • the discriminator 110 is generated by the generation unit 104 using the imaged data as the target of machine learning processing.
  • the analysis unit 106 uses the discriminator 110 to perform a state analysis process of the production facility 10.
  • the state of the production facility 10 determined by the discriminator 110 is, for example, normal or any other state.
  • states other than normal are also referred to as “Unknown”.
  • the discriminator 110 does not discriminate an abnormal state.
  • the machine learning process of the discriminator 110 will be described in detail in an embodiment described later.
  • the discrimination result of the discriminator 110 may be output so that the operator can refer to it.
  • the output method may be displayed on the display of the analysis device 100, may be printed out from the analysis device 100 on a printer, or may be transmitted to another device (for example, an operation terminal) via a communication line. May be done.
  • FIG. 7 is a flowchart showing an example of the operation of the analysis device 100 of this embodiment.
  • the image processing unit 102 visualizes the detection result of the vibration sensor provided in the production facility 10 (step S101).
  • the generation unit 104 generates the discriminator 110 by using the imaged data as the target of the machine learning process (step S103).
  • the analysis unit 106 uses the discriminator 110 to perform a state analysis process of the production facility 10 (step S105).
  • FIG. 8 is a flowchart showing an example of a detailed flow of the imaging process of step S101 of FIG.
  • the image processing unit 102 subjects the measurement data of the sensor 12 to FFT processing and frequency division (step S113), images the obtained vibration spectrum data, and outputs the image data (step S115).
  • the image data obtained in step S115 is subjected to state analysis processing of the production facility 10 using the discriminator 110 by the analysis unit 106 in step S103 of FIG.
  • the discriminator 110 targets the image data imaged by the generation unit 104 as the target of the machine learning process.
  • the analysis unit 106 uses the discriminator 110 to perform a state analysis process of the production facility 10.
  • the amount of data can be compressed by imaging the vibration waveform data that is the target of the machine-learning processing, so that the processing load is reduced.
  • the processing speed can be increased as well as the processing speed is reduced.
  • FIG. 9 is a functional block diagram showing a logical configuration of the analysis device 100 of this embodiment.
  • the analysis device 100 includes an image processing unit 102 similar to the analysis device 100 of FIG. 4, a generation unit 104, and an analysis unit 106, and further includes an abnormality determination unit 120.
  • the abnormality determination unit 120 uses the first discriminator 124 to perform abnormality determination processing of the production facility 10 based on the detection result of the vibration sensor provided in the production facility 10.
  • the abnormality determination unit 120 outputs the result of the abnormality determination process using the first discriminator 124.
  • the output result may be used for the monitoring process of the production facility 10 in the facility monitoring system 1.
  • the image processing unit 102 is that the detection result to be imaged is a detection result that the first discriminator 124 cannot determine whether the image processing unit 102 is normal or abnormal. Different from the part 102.
  • the generation unit 104 generates the second discriminator 126 for performing the machine learning process on the measurement data which cannot be discriminated as normal or abnormal by the threshold analysis.
  • the vibration measured by the vibration sensor contains multiple vibration waveforms consisting of multiple factors.
  • frequency analysis is performed on measurement data detected by a vibration sensor by performing a Fast Fourier Transform (FFT) process.
  • FFT Fast Fourier Transform
  • the characteristic frequency (peak) is detected, and the abnormality can be diagnosed by determining whether the detected peak level is normal or abnormal by the threshold value.
  • the generation unit 104 generates the second discriminator 126 by using the measurement data, which cannot be discriminated as normal or abnormal by the threshold analysis, as the target of the machine learning process.
  • the second discriminator 126 corresponds to the discriminator 110 in FIG.
  • the first discriminator 124 cannot discriminate means that, for example, when the vibration waveform pattern registered in the first discriminant model 128 and the detection result do not match, the likelihood of the matching result is equal to or more than a predetermined value. This is the case when the reliability is not obtained and the reliability higher than the standard required for equipment diagnosis (for example, the detection rate of 90% or more) is not obtained.
  • the analysis unit 106 uses the second discriminator 126 to perform a state analysis process of the production facility 10.
  • the analysis unit 106 uses the second discriminator 126 to analyze the state of the production facility 10 with respect to the data obtained by imaging the detection result in which the first discriminator 124 cannot discriminate between normal and abnormal.
  • the first discriminator 124 uses the first discriminant model 128 to discriminate between normal and abnormal.
  • the first discriminant model 128 is, for example, a model using pattern matching of vibration spectrum or threshold determination.
  • the first discriminator 124 uses the frequency distribution obtained by subjecting the measurement data of the vibration sensor to the FFT processing to determine the peak level of the specific frequency, the ratio of the maximum and average peak levels, and the S/N ratio (Signal-to -Noise ratio), and at least one of the integrated value of peak level in the range of the specific frequency, set the threshold value to the calculated value, and perform the abnormality determination process based on whether it is within the range of the threshold value. To do. If it is within the threshold range, it is determined to be normal, and if it is outside the threshold range, it is determined to be abnormal.
  • the first discriminator 124 may perform the above-described determination process using a plurality of specific frequencies. In that case, the first discriminator 124 may output the determination result for each specific frequency. Further, the first discriminator 124 may determine that there is an abnormality when even one of the plurality of specific frequencies is out of the threshold value, or may identify a predetermined number or more of the plurality of specific frequencies. It may be judged to be abnormal if there is a value outside the threshold value for the frequency, or may be judged to be abnormal if all of the multiple specific frequencies are outside the threshold value, or all of the multiple specific frequencies. May be determined to be normal when is within the threshold range.
  • a combination of a plurality of values of specific frequencies is modeled for each fault event and registered in the first discriminant model 128, and the first discriminator 124 performs the pattern matching with the first discriminant model 128 to cause a defect.
  • the event may be determined.
  • the abnormality determination unit 120 may notify the operator of the corresponding malfunction event item.
  • Threshold may be set automatically or may be set manually by an operator.
  • the generation unit 104 for the vibration characteristic pattern (frequency distribution) registered in the first discriminant model 128, the peak level of the specific frequency, the ratio of the maximum and average peak levels, and the S/N ratio ( Signal-to-Noise ratio) and the integrated value of the peak level in the range of the specific frequency, and the threshold value may be set by detecting the boundary range between the normal time and the abnormal time.
  • the generation unit 104 for the vibration characteristic pattern (frequency distribution) registered in the first discriminant model 128, the peak level of the specific frequency, the ratio of the maximum and average peak levels, and the S/N ratio (Signal). -to-Noise ratio) and at least one of the integrated value of the peak level in the range of the specific frequency are presented to the operator, and the operator sets the threshold value of the first discriminator 124. May be accepted respectively.
  • the result may be displayed on the display of the analysis device 100, may be printed out from the analysis device 100 on a printer, or may be output via a communication line. It may be transmitted to a device (for example, an operation terminal or the like).
  • the second discriminator 126 corresponds to the discriminator 110 generated by the generation unit 104 and generated by the generation unit 104 in the above embodiment.
  • the generator 104 machine-learns only the data discriminated as normal by the second discriminator 126, normalizes the normal state, and updates the second discriminant model 130.
  • the normalization is to model the pattern of the imaging data of the measurement data of the vibration sensor at the normal time.
  • the second discriminator 126 extracts data outside the range of the second discriminant model 130, for example, as “Unknown”. That is, with respect to the detection result which cannot be discriminated by the first discriminator 124 by the analysis unit 106 using the second discriminator 126, the state of the production facility 10 is further determined based on the second discriminant model 130. , Normal or "Unknown" is determined.
  • the second discriminator 126 does not discriminate an abnormal state of the production facility 10, and discriminates a state other than normal as “Unknown”.
  • the data discriminated as “Unknown” by the discriminator 110 may be referred to and analyzed by the operator.
  • the result analyzed by the operator may be reflected in the threshold value of the abnormality determination process of the detection result.
  • a method of using the data determined as “Unknown” will be described in detail in an embodiment described later.
  • FIG. 10 is a flowchart showing an example of the operation of the analysis device 100 of this embodiment.
  • the abnormality determination unit 120 uses the first discriminator 124 to perform abnormality determination processing of the production facility 10 based on the detection result of the vibration sensor provided in the production facility 10 (step S121).
  • the image processing unit 102 visualizes the detection result of which the first discriminator 124 could not determine whether it is normal or abnormal (NO in step S123) (step S125). If it can be determined (YES in step S123), this process ends.
  • the generation unit 104 generates the second discriminator 126 by using the imaged data as the target of the machine learning process (step S127).
  • the analysis unit 106 uses the second discriminator 126 to perform a state analysis process of the production facility 10 (step S129).
  • the computer program of the present embodiment causes the computer (processor 50 in FIG. 3) for realizing the analysis apparatus 100 to detect an abnormality in the production facility 10 based on the detection result of the vibration sensor 12 provided in the production facility 10.
  • a procedure for performing the determination process using the first discriminator 124, a procedure for imaging the detection result in which the first discriminator 124 cannot determine whether it is normal or abnormal, and the imaged data is the target of the machine learning process. It is described that the procedure for generating the second discriminator 126 and the procedure for performing the state analysis process of the production facility 10 by using the second discriminator 126 are executed.
  • the abnormality determination unit 120 uses the first discriminator 124 to obtain no determination result in the abnormality determination process based on the detection result of the vibration sensor, the vibration at that time is detected.
  • the detection result of the sensor is imaged by the image processing unit 102, and the second discriminator 126 discriminates whether it is normal or “Unknown”.
  • the second discriminator 126 performs machine learning only on the detection result of the vibration sensor corresponding to the normal state and normalizes the normal state. Even in the case where the manufacturing conditions are always in fluidity such as in production, and it is difficult to accumulate the information on the failure event, the accuracy of the determination can be improved.
  • the amount of data can be compressed by imaging the vibration waveform data that is the target of the machine-learning process, so that the processing load is reduced. At the same time, the processing speed can be increased.
  • FIG. 11 is a flowchart showing an example of a procedure of abnormality determination processing in the analysis device 100 of this embodiment.
  • the present embodiment is the same as the analysis apparatus 100 in FIG. 9 except that it has a configuration for outputting the result of the abnormality determination processing of the production facility 10 using the first discriminator 124.
  • the abnormality determination unit 120 acquires vibration data from the vibration sensor provided in the production facility 10 (step S301). Then, the abnormality determination unit 120 executes the abnormality determination process of the vibration data acquired using the first discriminator 124 (step S303). Here, after the vibration data is subjected to FFT processing, the first discriminant model 128 of the abnormality determination unit 120 determines whether the vibration data is normal or abnormal (step S305).
  • the abnormality determination unit 120 determines the peak level of the specific frequency, the ratio of the maximum and average peak levels, the S/N ratio (Signal-to-Noise ratio), and the range of the specific frequency. For each of at least any one of the integrated values of the peak levels, the threshold value is used to determine whether these values are normal or abnormal (step S305). If it is within the threshold range (NO in step S305), the first discriminator 124 determines that the production facility 10 is normal (step S307), and the abnormality determination unit 120 determines that the production facility 10 is normal. Is output (step S311).
  • the first discriminator 124 determines that the production facility 10 is in an abnormal state (step S309), and the determination result indicating that the production facility 10 is in an abnormal state. Is output (step S311).
  • step S305 If the first discriminator 124 cannot determine whether it is normal or abnormal (determination in step S305 is not possible), the abnormality determination unit 120 outputs vibration data to the image processing unit 102 (step S313), and It progresses to step S125 of 10.
  • the first discriminator 124 is used for the abnormality determination processing of the production facility 10, and the second discriminator 126 is not used for the abnormality determination processing.
  • the discrimination result of the second discriminator 126 is used to update the second discriminant model 130 of the first discriminator 124, and this configuration will be described in an embodiment described later.
  • the abnormality determination unit 120 performs the abnormality determination processing of the production facility 10 using the first discriminator 124, and the abnormality determination unit 120 causes the second discriminator 126 to operate the production facility 10. It is not used for the abnormality determination processing of.
  • the first discriminator 124 can be updated using the discrimination result of the second discriminator 126 updated by the generation unit 104. According to this configuration, since the abnormality determination process can be performed using the first discriminator 124 that reflects the discrimination result of the second discriminator 126, the accuracy of the determination result can be improved.
  • FIG. 12 is a functional block diagram showing a logical configuration of the image processing unit 102 of the analysis device 100 of this embodiment.
  • the analysis apparatus 100 of the present embodiment is the same as the above-described embodiment except that the image processing unit 102 has a configuration of performing noise removal from measurement data and then performing imaging processing.
  • the configuration of this embodiment may be combined with the configuration of any other embodiment.
  • the image processing unit 102 includes a noise removal unit 112 and an imaging processing unit 114.
  • the noise removal unit 112 performs noise removal processing on the detection result.
  • the imaging processing unit 114 images the detection result after the noise removal processing by the noise removal unit 112 is performed. By removing noise, the vibration waveform of measurement data becomes clear.
  • the noise removal processing includes, for example, measuring and storing environmental noise from a plurality of arranged vibration sensors in advance and performing difference processing based on the noise data.
  • the noise removal process may be performed by another method.
  • FIG. 13 is a flowchart showing an example of the operation of the image processing unit 102 of the analysis device 100 of this embodiment.
  • the flowchart of FIG. 13 includes step S111 in addition to steps S113 and S115 of the flowchart of FIG. 8 described in the above embodiment.
  • step S111 the noise removal unit 112 performs noise removal processing on measurement data.
  • the image processing unit 114 performs FFT processing on the noise-removed data in step S111 and frequency division processing (step S113), and the obtained vibration spectrum data is imaged to output the image data (step S115).
  • the image data obtained in step S115 is subjected to state analysis processing of the production facility 10 using the discriminator 110 by the analysis unit 106 in step S103 of FIG.
  • the measurement data of the sensor 12 that has been subjected to the noise removal processing by the noise removal unit 112 is imaged by the imaging processing unit 114.
  • the vibration waveform of the measurement data becomes clear by the noise removal, and the FFT processing and The accuracy of the imaging process is improved, and the accuracy and reliability of the measurement data analysis result are improved.
  • FIG. 14 is a functional block diagram showing a logical configuration of the analysis device 100 of this embodiment.
  • the analysis device 100 of the present embodiment is different from the above-described embodiment in that it extracts data that is determined to be abnormal by the state analysis process of the production facility 10, and based on the data, the first first discriminator 124 of the first discriminator 124.
  • the configuration is the same as that of the above embodiment except that the discriminant model 128 is updated.
  • the analysis device 100 includes an image processing unit 102 similar to the analysis device 100 of FIG. 9, a generation unit 104, an analysis unit 106, and an abnormality determination unit 120, and further includes an extraction unit 140.
  • the extraction unit 140 extracts data that is determined to be abnormal (“Unknown”) in the state analysis processing using the second discriminator 126.
  • the generator 104 receives the correction information based on the extracted data and updates the first discriminant model 128 of the first discriminator 124.
  • the correction information includes information in which the malfunction event and the vibration characteristic are associated with each other.
  • the extracted data means the raw vibration waveform data received from the vibration sensor before the image processing by the image processing unit 102 and the time information of the data, which corresponds to the imaged data determined as “Unknown”. including.
  • the vibration indicated by the vibration data determined to be “Unknown” may be caused by a failure event of the production facility 10 that has not been specified yet.
  • the operator manually analyzes the extracted vibration data together with the time information by using the operation information, the status information, the information of the workpiece, etc. of the production equipment 10 that the equipment monitoring system 1 has, and the vibration concerned. Identify the failure event that caused the.
  • the analysis device 100 outputs the data extracted by the extraction unit 140 and presents it to the operator.
  • the data may be displayed on the display of the analysis device 100, may be printed out from the analysis device 100 on a printer, or may be output via a communication line. It may be transmitted to a device (for example, an operation terminal or the like).
  • the operator inputs the correction information 30 of FIG. 15 in which the vibration characteristic information of the corresponding vibration is associated with the defect information specified by the operator to the analysis device 100 using an operation screen or the like.
  • the generation unit 104 receives the input correction information 30 and updates the first discriminant model 128 of the first discriminator 124.
  • FIG. 16 is a flowchart showing an example of a detailed flow of the discrimination processing by the analysis unit 106.
  • the analysis unit 106 applies the image data output in step S115 of FIG. 13 to the second discriminator 126 to perform a state analysis process of the production facility 10 (step S401).
  • the data determined to be normal by the second discriminator 126 (“normal” in step S401) is passed to the generation unit 104, and machine learning processing is performed as normal data (step S403).
  • the analysis unit 106 extracts the data outside the normalization range by the second discriminator 126 (“Unknown” in step S401) (step S405).
  • step S115 of FIG. 8 may be similarly processed using the flow of FIG.
  • FIG. 17 is a flowchart showing an example of a discriminator update processing procedure using the data extracted in step S405 of FIG.
  • the extraction unit 140 outputs the data (vibration data and time information) extracted in step S405 (step S411).
  • the operator analyzes the vibration data displayed in step S411 together with the time information using the operation information of the production facility 10, the state information, the information of the workpiece, etc., which the facility monitoring system 1 has, Identify the faulty event that caused the vibration.
  • the operator creates information in which the identified malfunction event and the corresponding vibration characteristic information are associated with each other, and inputs the information as correction information for the threshold of the first discriminator 124 according to the operation screen of the analysis device 100.
  • the generation unit 104 receives the input correction information (step S413), and updates the first discriminator 124 using the received correction information (step S415).
  • the vibration characteristic information and the malfunction event included in the received correction information are registered in the first discrimination model 128, and the peak level of the specific frequency, the ratio of the maximum and average peak levels, and the S/N ratio ( Signal-to-Noise ratio) and the integrated value of the peak level in the range of the specific frequency, and calculate the boundary range between normal and abnormal conditions and set the threshold.
  • the operator may set thresholds based on the respective values obtained from the vibration characteristic information corresponding to the identified malfunction event, and input the thresholds using the operation screen.
  • the extracting unit 140 extracts the “Unknown” data of the second discriminator 126, presents it to the operator, and analyzes the operator by the generating unit 104.
  • the correction information associated with and is received, and the first discriminator 124 is updated based on the correction information.
  • the measurement data that the first discriminator 124 cannot discriminate is further discriminated by the second discriminator 126, and the data that is “Unknown” is extracted. Since the operator analyzes and reflects the result in the first discriminator 124, the accuracy of the abnormality determination processing can be improved.
  • the second discriminator 126 machine-learns only the information in the normal state, it may be the case that the manufacturing condition is always fluid and the information on the defective event is difficult to be accumulated in a small-lot, large-variety product or variable production.
  • the first determination model 128 can be updated, the accuracy of determining the abnormal state of the production facility 10 can be improved.
  • the analysis apparatus 100 of FIG. 14 has a configuration in which the extraction unit 140 is provided in the configuration of the analysis apparatus 100 of FIG. As a modification thereof, the analysis device 100 of FIG. 4 may be provided with the extraction unit 140.
  • the analysis device 100 further includes an abnormality determination unit 120 and an extraction unit 140.
  • the abnormality determination unit 120 performs the characteristic analysis process of the vibration of the vibration sensor indicated by the detection result, and performs the abnormality determination process of the production facility 10 using the threshold value.
  • the extraction unit 140 extracts data that is determined by the state analysis process that the production facility 10 is not in a normal state.
  • the generation unit 104 receives the correction information based on the extracted data and updates the threshold value.
  • the correction information includes information in which the fault event and the vibration characteristic are associated with each other.
  • FIG. 18 is a flowchart showing an example of a processing procedure when it is determined as normal in step S401 of the state analysis processing of FIG.
  • This embodiment is the same as the above embodiment except that the detection result determined by the second discriminator 126 is normal is used as the teacher data of the second discriminator 126.
  • the generation unit 104 uses the detection result determined to be normal by the state analysis processing using the second discriminator 126 (normal in step S401 of FIG. 16) as the second teacher data in the normal state of the production facility 10.
  • the second discriminant model 130 of the discriminator 126 is updated (step S501).
  • the second discriminant model 130 is updated by the generation unit 104 with the detection result determined as normal by the state analysis processing using the second discriminator 126 as the teacher data of the normal state of the production facility 10. To be done.
  • the teacher data in the normal state can be generated from the measurement data that could not be discriminated in the abnormality discriminating process by the first discriminator 124, and the second discriminant model 130 can be updated. It is possible to improve the determination accuracy of.
  • the analysis device 100 may include a third discriminator (not shown).
  • FIG. 19 is a flowchart showing an example of a processing procedure in which the analysis device 100 constructs a defect model using the third discriminator and identifies a defect event.
  • the third discriminator acquires the measurement data discriminated as normal by the first discriminator 124 (step S601). Furthermore, the third discriminator acquires the measurement data discriminated as Unknown by the second discriminator 126 (step S603). Then, the third discriminator machine-learns these data and constructs a model in which defective events are classified (step S605).
  • the noise removal processing and the imaging processing described in the above embodiment may be performed on each measurement data machine-learned by the third discriminator.
  • step S607 it is determined whether the measurement data of the sensor 12 of the production facility 10 is normal or abnormal, and if it is abnormal, the defective event is identified.
  • the position information of each of the plurality of sensors 12 is stored in the facility information 24, and the relationship between the vibration data and the position information is further machine-learned, and the classification model 202, the first discrimination model 128, and the second It may be reflected in at least one of the discrimination models 130.
  • information such as measurement conditions (equipment type, environment (temperature, humidity), etc.) of measurement data of a plurality of sensors 12 may be stored in the equipment information 24.
  • the measurement data is grouped with the measurement data of conditions close to the measurement condition, the operation condition included in the operation information of the production facility 10, the operation condition, etc., and the measurement data is machine-learned for each group, and the classification model 202 and the first discrimination model 128 are used.
  • the second discriminant model 130 may be reflected.
  • FIG. 20 is a flow chart for explaining the analyzing apparatus of the first embodiment.
  • the abnormality determination unit 120 performs the FFT processing in the first discriminator 124 (step S11).
  • the vibration characteristic is specified by the pattern matching process using the first discrimination model 128.
  • the first discriminator 124 determines that the vibration characteristics are normal when the vibration characteristics are within the threshold range, and determines that the vibration characteristics are abnormal when the vibration characteristics are outside the threshold range.
  • the abnormality determination unit 120 outputs this result to the equipment monitoring system 1 as an abnormality determination result of the production equipment 10 (not shown).
  • the abnormality determination unit 120 extracts the data for which the first discriminator 124 could not discriminate whether it is normal in step S11 (step S13), and sets it as the target of the machine learning process of the second discriminator 126. It is passed to the analysis unit 106.
  • the analysis unit 106 performs noise removal processing on the measurement data of the sensor 12 that cannot be discriminated by the first discriminator 124 (step S15), frequency-analyzes it, and forms an image (step S17).
  • the analysis unit 106 determines the imaged data using the second discriminator 126 (step S19), and determines whether the state of the production facility 10 is normal (step S21). When it is determined to be normal (YES in step S21), the generation unit 104 machine-learns the measurement data determined to be normal, and updates the second discriminant model 130 of the second discriminator 126 (step S23). ..
  • the extraction unit 140 extracts and outputs the unknown measurement data (step S31).
  • the operator refers to the extracted unknown data and analyzes it together with the operation information of the production facility 10 and the like, and identifies the defective phenomenon. Then, the correction information in which the fault event and the vibration characteristic are associated with each other is input to the analysis device 100 (step S33).
  • the generation unit 104 receives the input correction information, and updates the first discrimination model 128 and the threshold value based on the received correction information (step S35).
  • the analyzer 100 causes the second discriminator 126 to machine-learn the measurement data that could not be abnormally determined by the first discriminator 124, thereby extracting abnormal data that is not normal and producing the production facility.
  • the operator analyzes the operation information together with the operation information of 10 to identify a defective event, and inputs the corrected information to the analysis apparatus 100 as the correction information in which the vibration characteristic and the defective event are associated with each other.
  • the first discriminant model 128 and the threshold can be updated.
  • FIG. 21 is a flow chart for explaining the analyzing apparatus of the second embodiment.
  • the analysis apparatus 100 of this embodiment includes a third discriminator 200 in addition to the first discriminator 124 and the second discriminator 126.
  • the third discriminator 200 performs machine learning using the measurement data discriminated as normal by the first discriminator 124 and the measurement data discriminated as Unknown by the second discriminator 126 (step S41).
  • the second discriminator 126 constructs a normal and abnormal classification model 202 by machine learning.
  • the classification model 202 classifies defective events into classes.
  • the third discriminator 200 can discriminate whether the measurement data is normal or abnormal, and can further discriminate and specify the faulty event.
  • Image processing means for imaging the detection result of the vibration sensor provided in the production facility, Generating means for generating a discriminator by subjecting the imaged data to machine learning processing; And an analysis unit that performs a state analysis process of the production facility using the discriminator. 2.
  • a determination unit that performs a characteristic analysis process of the vibration of the vibration sensor indicated by the detection result, and performs an abnormality determination process of the production facility using a threshold value
  • the generation means receives correction information based on the extracted data, updates the threshold value,
  • the said correction information is an analysis apparatus containing the information which linked
  • Determination means for performing abnormality determination processing of the production facility using the first discriminator based on the detection result of the vibration sensor provided in the production facility; Image processing means for imaging the detection result, which cannot be discriminated as normal or abnormal by the first discriminator; Generating means for generating a second discriminator using the imaged data as a target of machine learning processing; Analysis means for performing a state analysis process of the production facility using the second discriminator; With Analyzer. 4. 3. In the analyzer described in The state analysis process further comprises an extracting unit for extracting data determined to be in a non-normal state of the production facility, The generation means receives correction information based on the extracted data, updates the first discriminator, The said correction information is an analysis apparatus containing the information which linked
  • the generation unit uses the data determined to be normal by the state analysis process as teacher data for the machine learning process.
  • the generation unit uses the data determined to be normal by the state analysis process as teacher data for the machine learning process.
  • 6. 1. To 5.
  • the analysis device according to any one of Further comprising processing means for performing noise removal processing on the detection result, The analysis device, wherein the image processing means images the detection result after the noise removal processing is performed by the processing means. 7. 1. To 6.
  • the analysis device according to any one of The production facility is a belt conveyor,
  • the said vibration sensor is an analysis apparatus which is a some vibration sensor provided in the said belt conveyor.
  • the analysis device Visualize the detection result of the vibration sensor installed in the production facility, Generate a discriminator with the imaged data as the target of machine learning processing, Performing a state analysis process of the production facility using the discriminator, analysis method. 9. 8. In the analysis method described in The analysis device further comprises Performing a characteristic analysis process of the vibration of the vibration sensor indicated by the detection result, performing an abnormality determination process of the production facility using a threshold, Extracting the data determined that the production equipment is not in a normal state in the state analysis process, Accept correction information based on the extracted data, update the threshold, The said correction information is an analysis method containing the information which linked
  • the analysis device Based on the detection result of the vibration sensor provided in the production equipment, the abnormality determination processing of the production equipment is performed using the first discriminator, The detection result, which could not be discriminated as normal or abnormal by the first discriminator, is imaged, A second discriminator is generated by using the imaged data as a target of machine learning processing, Performing a state analysis process of the production facility using the second discriminator, analysis method. 11. 10. In the analysis method described in The analysis device further comprises Extracting the data determined that the production equipment is not in a normal state in the state analysis process, Accepting correction information based on the extracted data, updating the first discriminator, The said correction information is an analysis method containing the information which linked
  • the analysis device further comprises An analysis method, wherein the data determined to be normal by the state analysis process is used as teacher data for the machine learning process. 13. 8. To 12. In the analysis method described in any one of The analysis device further comprises Noise removal processing is performed on the detection result, An analysis method of imaging the detection result after the noise removal processing is performed. 14. 8. To 13. In the analysis method described in any one of The production facility is a belt conveyor, An analysis method, wherein the vibration sensor is a plurality of vibration sensors provided on the belt conveyor.
  • Procedure for imaging the detection result of the vibration sensor installed in the production facility A procedure for generating a discriminator by subjecting the imaged data to machine learning processing, A program for executing a procedure of performing abnormality determination processing of the production facility using the discriminator. 16. 15.
  • a procedure for performing an abnormality determination process of the production facility using a threshold A procedure for extracting data that is determined in the state analysis process that the production facility is not in a normal state,
  • the correction information based on the extracted data is accepted, and the procedure of updating the threshold value is further executed by a computer,
  • the said correction information is a program containing the information which linked
  • a procedure for performing noise removal processing on the detection result A program for causing a computer to further execute the step of imaging the detection result after the noise removal processing is performed in the step of imaging. 21. 15.
  • the production facility is a belt conveyor, A program in which the vibration sensor is a plurality of vibration sensors provided on the belt conveyor.

Landscapes

  • Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Software Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Computation (AREA)
  • Medical Informatics (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Physics (AREA)
  • Artificial Intelligence (AREA)
  • Multimedia (AREA)
  • Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
  • Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)

Abstract

An analysis device (100) is provided with: an image processing unit (102) that converts, into an image, a the result of detection by a vibration sensor provided to a production facility (10); a generation unit (104) that generates a discriminator (110) by subjecting the imaged data to a machine learning process; and an analysis unit (106) that performs a state analysis process of the production facility (10) by using the discriminator (110).

Description

解析装置、解析方法、およびプログラムAnalysis device, analysis method, and program

　本発明は、解析装置、解析方法、およびプログラムに関し、特に、設備の状態を監視するセンサのデータを解析する解析装置、解析方法、およびプログラム関する。 The present invention relates to an analysis device, an analysis method, and a program, and more particularly to an analysis device, an analysis method, and a program for analyzing data of a sensor that monitors the state of equipment.

　機械系設備を用いた生産材の製造品質管理に振動や音響センサを用いて状態監視する方法がある。例えば、生産材の加工時に加工機で生じた振動データを取得し、異常振動を捉えたときに生産材の加工をストップすることにより製造ロスを回避することや、生産設備の稼働状況を振動にて監視し、メンテナンスの効率化や、設備を長寿命化するための最適な稼働条件を見出すことなどにより、製造業の生産効率を向上させる技術として注目されている。 There is a method to monitor the condition using vibration and acoustic sensors for manufacturing quality control of production materials using mechanical equipment. For example, by acquiring vibration data generated by a processing machine during processing of production materials and avoiding production loss by stopping the processing of production materials when abnormal vibrations are captured, the operating status of production equipment can be used as vibration. It is attracting attention as a technology that improves the production efficiency of the manufacturing industry by monitoring and monitoring it, improving the efficiency of maintenance, and finding the optimal operating conditions for extending the life of the equipment.

　特許文献１には、監視対象の設備にセンサを取り付け、当該センサが測定した時系列データに基づきその設備の監視を行う方法が開示されている。 Patent Document 1 discloses a method in which a sensor is attached to a facility to be monitored and the facility is monitored based on time series data measured by the sensor.

特開２００９－２７０８４３号公報JP, 2009-270843, A

　上述した特許文献１に記載されたシステムのように、センサで収集したデータに基づき設備の異常等を検出しようとすると、解析すべきデータの量が多くなり得る。また、ディープラーニングなどの機械学習により正常／異常状態モデルを構築することも考えられるが、この方法では、あくまで異常と推察される状態を抽出させるだけに留まり、稼働を停止するなどのアクションにしか適用できない。 Like the system described in Patent Document 1 described above, when an abnormality of equipment is detected based on the data collected by the sensor, the amount of data to be analyzed may increase. It is also possible to build a normal/abnormal state model by machine learning such as deep learning, but this method only extracts the state that is suspected to be abnormal and only takes action such as stopping the operation. Not applicable.

　本発明は上記事情に鑑みてなされたものであり、その目的とするところは、生産設備の状態監視を効率よく高精度に行う技術を提供することにある。 The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a technique for efficiently and accurately monitoring the state of production equipment.

　本発明の各側面では、上述した課題を解決するために、それぞれ以下の構成を採用する。 In each aspect of the present invention, the following configurations are adopted to solve the above problems.

　第一の側面は、解析装置に関する。
　第一の側面に係る第１の解析装置は、
　生産設備に設けられた振動センサの検出結果を画像化する画像処理手段と、
　前記画像化したデータを機械学習処理の対象にして判別器を生成する生成手段と、
　前記判別器を用いて前記生産設備の状態解析処理を行う解析手段と、を有する。
　第一の側面に係る第２の解析装置は、
　生産設備に設けられた振動センサの検出結果に基づいて当該生産設備の異常判定処理を第１の判別器を用いて行う判定手段と、
　前記第１の判別器により正常か異常かの判別ができなかった前記検出結果を画像化する画像処理手段と、
　前記画像化したデータを機械学習処理の対象として第２の判別器を生成する生成手段と、
　前記第２の判別器を用いて前記生産設備の状態解析処理を行う解析手段と、を有する。 The first aspect relates to an analysis device.
The first analysis device according to the first aspect is
Image processing means for imaging the detection result of the vibration sensor provided in the production facility,
Generating means for generating a discriminator by subjecting the imaged data to machine learning processing;
And an analysis unit that performs a state analysis process of the production facility using the discriminator.
The second analysis device according to the first aspect,
Determination means for performing abnormality determination processing of the production facility using the first discriminator based on the detection result of the vibration sensor provided in the production facility;
Image processing means for imaging the detection result, which cannot be discriminated as normal or abnormal by the first discriminator;
Generating means for generating a second discriminator using the imaged data as a target of machine learning processing;
Analysis means for performing a state analysis process of the production facility using the second discriminator.

　第二の側面は、少なくとも１つのコンピュータにより実行される解析方法に関する。
　第二の側面に係る第１の解析方法は、
　解析装置が、
　生産設備に設けられた振動センサの検出結果を画像化し、
　前記画像化したデータを機械学習処理の対象にして判別器を生成し、
　前記判別器を用いて前記生産設備の状態解析処理を行う、ことを含む。
　第二の側面に係る第２の解析方法は、
　解析装置が、
　生産設備に設けられた振動センサの検出結果に基づいて当該生産設備の異常判定処理を第１の判別器を用いて行い、
　前記第１の判別器により正常か異常かの判別ができなかった前記検出結果を画像化し、
　前記画像化したデータを機械学習処理の対象として第２の判別器を生成し、
　前記第２の判別器を用いて前記生産設備の状態解析処理を行う、ことを含む。 The second aspect relates to at least one computer implemented analysis method.
The first analysis method according to the second aspect is
The analysis device
Visualize the detection result of the vibration sensor installed in the production facility,
Generate a discriminator with the imaged data as the target of machine learning processing,
Performing state analysis processing of the production facility using the discriminator.
The second analysis method according to the second aspect is
The analysis device
Based on the detection result of the vibration sensor provided in the production equipment, the abnormality determination processing of the production equipment is performed using the first discriminator,
The detection result, which could not be discriminated as normal or abnormal by the first discriminator, is imaged,
A second discriminator is generated by using the imaged data as a target of machine learning processing,
Performing state analysis processing of the production facility using the second discriminator.

　なお、本発明の他の側面としては、上記第二の側面の方法を少なくとも１つのコンピュータに実行させるプログラムであってもよいし、このようなプログラムを記録したコンピュータが読み取り可能な記録媒体であってもよい。この記録媒体は、非一時的な有形の媒体を含む。
　このコンピュータプログラムは、コンピュータにより実行されたとき、コンピュータに、解析装置上で、その解析方法を実施させるコンピュータプログラムコードを含む。 As another aspect of the present invention, it may be a program that causes at least one computer to execute the method of the second aspect, or a computer-readable recording medium that records such a program. May be. The recording medium includes a non-transitory tangible medium.
The computer program includes a computer program code that, when executed by a computer, causes the computer to perform the analysis method on an analysis device.

　なお、以上の構成要素の任意の組合せ、本発明の表現を方法、装置、システム、記録媒体、コンピュータプログラムなどの間で変換したものもまた、本発明の態様として有効である。 It should be noted that any combination of the above constituent elements and the expression of the present invention converted between a method, a device, a system, a recording medium, a computer program, etc. are also effective as an aspect of the present invention.

　また、本発明の各種の構成要素は、必ずしも個々に独立した存在である必要はなく、複数の構成要素が一個の部材として形成されていること、一つの構成要素が複数の部材で形成されていること、ある構成要素が他の構成要素の一部であること、ある構成要素の一部と他の構成要素の一部とが重複していること、等でもよい。 Further, the various constituent elements of the present invention do not necessarily have to be independently present, and a plurality of constituent elements are formed as one member, and one constituent element is formed by a plurality of members. May be present, a certain component may be a part of another component, a part of a certain component may overlap a part of another component, and the like.

　また、本発明の方法およびコンピュータプログラムには複数の手順を順番に記載してあるが、その記載の順番は複数の手順を実行する順番を限定するものではない。このため、本発明の方法およびコンピュータプログラムを実施するときには、その複数の手順の順番は内容的に支障のない範囲で変更することができる。 Moreover, although the method and the computer program of the present invention describe a plurality of procedures in order, the order of description does not limit the order in which the plurality of procedures are executed. Therefore, when carrying out the method and computer program of the present invention, the order of the plurality of procedures can be changed within a range that does not hinder the contents.

　さらに、本発明の方法およびコンピュータプログラムの複数の手順は個々に相違するタイミングで実行されることに限定されない。このため、ある手順の実行中に他の手順が発生すること、ある手順の実行タイミングと他の手順の実行タイミングとの一部ないし全部が重複していること、等でもよい。 Furthermore, the plurality of procedures of the method and computer program of the present invention are not limited to being executed at different timings. For this reason, another procedure may occur during the execution of a certain procedure, the execution timing of a certain procedure and the execution timing of another procedure may partially or entirely overlap, and the like.

　上記各側面によれば、生産設備の状態監視を効率よく高精度に行う技術を提供することができる。 According to each of the above aspects, it is possible to provide a technology for efficiently and accurately monitoring the state of production equipment.

　上述した目的、およびその他の目的、特徴および利点は、以下に述べる好適な実施の形態、およびそれに付随する以下の図面によってさらに明らかになる。 The above-mentioned object, other objects, features and advantages will be further clarified by the preferred embodiments described below and the following drawings accompanying it.

本発明の実施の形態に係る解析装置を用いた設備監視システムのシステム構成を概念的に示す図である。It is a figure which shows notionally the system configuration of the equipment monitoring system using the analysis apparatus which concerns on embodiment of this invention. 本実施形態の記憶装置が記憶する振動データと設備情報のデータ構造の一例を示す図である。It is a figure which shows an example of the data structure of the vibration data and equipment information which the memory|storage device of this embodiment memorize|stores. 本実施形態の各装置のハードウェア構成を例示するブロック図である。It is a block diagram which illustrates the hardware constitutions of each apparatus of this embodiment. 本実施形態の解析装置の論理的な構成を示す機能ブロック図である。It is a functional block diagram which shows the logical structure of the analysis apparatus of this embodiment. は画像化前の測定データの時系列データを示す図である。FIG. 4 is a diagram showing time series data of measurement data before imaging. 画像処理部により画像化した画像データを示す図である。It is a figure which shows the image data imaged by the image processing part. 本実施形態の解析装置の動作の一例を示すフローチャートである。It is a flowchart which shows an example of operation|movement of the analysis apparatus of this embodiment. 図７のステップＳ１０１の画像化処理の詳細フローの一例を示すフローチャートである。8 is a flowchart showing an example of a detailed flow of an imaging process in step S101 of FIG. 本実施形態の解析装置の論理的な構成を示す機能ブロック図である。It is a functional block diagram which shows the logical structure of the analysis apparatus of this embodiment. 本実施形態の解析装置の動作の一例を示すフローチャートである。It is a flowchart which shows an example of operation|movement of the analysis apparatus of this embodiment. 本実施形態の解析装置における異常判定処理の手順の一例を示すフローチャートである。It is a flow chart which shows an example of the procedure of the abnormality judging processing in the analysis device of this embodiment. 本実施形態の解析装置の画像処理部の論理的な構成を示す機能ブロック図である。It is a functional block diagram which shows the logical structure of the image processing part of the analysis apparatus of this embodiment. 本実施形態の解析装置の画像処理部の動作の一例を示すフローチャートである。It is a flow chart which shows an example of operation of an image processing part of an analysis device of this embodiment. 本実施形態の解析装置の論理的な構成を示す機能ブロック図である。It is a functional block diagram which shows the logical structure of the analysis apparatus of this embodiment. 補正情報３０のデータ構造の一例を示す図である。It is a figure which shows an example of the data structure of the correction information 30. 解析部による判別処理の詳細フローの一例を示すフローチャートである。It is a flow chart which shows an example of a detailed flow of discrimination processing by an analysis part. 図１６のステップＳ４０６で抽出されたデータを用いた判別器の更新処理手順の一例を示すフローチャートである。17 is a flowchart illustrating an example of a discriminator update processing procedure using the data extracted in step S406 of FIG. 16. 図１６の状態解析処理のステップＳ４０１で正常と判別された場合の処理手順の一例を示すフローチャートである。17 is a flowchart showing an example of a processing procedure when it is determined to be normal in step S401 of the state analysis processing of FIG. 解析装置の不具合モデル構築処理の手順の一例を示すフローチャートである。It is a flow chart which shows an example of the procedure of defective model construction processing of an analysis device. 実施例１の解析装置を説明するためのフロー図である。FIG. 3 is a flowchart for explaining the analysis device of the first embodiment. 実施例２の解析装置を説明するためのフロー図である。FIG. 6 is a flowchart for explaining the analysis device of the second embodiment.

　以下、本発明の実施の形態について、図面を用いて説明する。尚、すべての図面において、同様な構成要素には同様の符号を付し、適宜説明を省略する。 Embodiments of the present invention will be described below with reference to the drawings. In all the drawings, the same constituents will be referred to with the same numerals, and the description thereof will not be repeated.

（第１の実施の形態）
　図１は、本発明の実施の形態に係る解析装置を用いた設備監視システム１のシステム構成を概念的に示す図である。
　設備監視システム１が監視対象とする設備は、生産設備１０であり、本実施形態では、ベルトコンベアを例として説明する。図の例では、ベルトコンベアを監視するための複数のセンサ１２がベルトコンベアの移動方向に沿って複数箇所に設置されている。各センサ１２は、例えば、振動センサである。また、一方向の振動を検知する振動センサの場合、複数方向の振動を検出するために１箇所につき複数の振動センサを設置してもよい。 (First embodiment)
FIG. 1 is a diagram conceptually showing a system configuration of an equipment monitoring system 1 using an analysis device according to an embodiment of the present invention.
The equipment monitored by the equipment monitoring system 1 is the production equipment 10, and in the present embodiment, a belt conveyor will be described as an example. In the illustrated example, a plurality of sensors 12 for monitoring the belt conveyor are installed at a plurality of locations along the moving direction of the belt conveyor. Each sensor 12 is, for example, a vibration sensor. Further, in the case of a vibration sensor that detects vibration in one direction, a plurality of vibration sensors may be installed at one location in order to detect vibration in multiple directions.

　振動センサから出力される測定データは、振動波形を示す時系列データである。
　振動センサは、監視対象の生産設備１０に生じた振動を測定する。振動センサは、一軸方向の加速度を測定する一軸加速度センサであってもよいし、三軸方向の加速度を測定する三軸加速度センサであってもよいし、その他であってもよい。なお、複数の振動センサは、同種の振動センサであってもよいし、複数種類の振動センサが混在してもよい。 The measurement data output from the vibration sensor is time series data indicating a vibration waveform.
The vibration sensor measures the vibration generated in the production facility 10 to be monitored. The vibration sensor may be a uniaxial acceleration sensor that measures acceleration in the uniaxial direction, a triaxial acceleration sensor that measures acceleration in the triaxial directions, or any other type. The plurality of vibration sensors may be the same kind of vibration sensor, or a plurality of kinds of vibration sensors may be mixed.

　解析装置１００は、ネットワーク３を介してＧＷ（GateWay）５と接続され、生産設備１０に複数設けられているセンサ１２から検出結果を受信する。解析装置１００は、記憶装置２０に接続される。記憶装置２０は、解析装置１００が解析する振動データが格納される。記憶装置２０は、解析装置１００とは別体の装置であってもよいし、解析装置１００の内部に含まれる装置であってもよいし、これらの組み合わせであってもよい。 The analysis device 100 is connected to the GW (GateWay) 5 via the network 3 and receives detection results from the plurality of sensors 12 provided in the production facility 10. The analysis device 100 is connected to the storage device 20. The storage device 20 stores vibration data analyzed by the analysis device 100. The storage device 20 may be a device separate from the analysis device 100, a device included in the analysis device 100, or a combination thereof.

　実施形態の各図において、本発明の本質に関わらない部分の構成については省略してあり、図示されていない。 In each of the drawings of the embodiments, the configuration of the portion not related to the essence of the present invention is omitted and not shown.

　図２は、本実施形態の記憶装置２０が記憶する振動データ２２と設備情報２４のデータ構造の一例を示す図である。
　振動データ２２は、振動センサを識別するセンサＩＤ毎に、時刻情報と、測定データとが紐付けられている。設備情報２４は、設備を識別する設備ＩＤ毎に、少なくとも一つの振動センサのセンサＩＤが紐付けられている。 FIG. 2 is a diagram showing an example of a data structure of the vibration data 22 and the facility information 24 stored in the storage device 20 of this embodiment.
The vibration data 22 is associated with time information and measurement data for each sensor ID that identifies the vibration sensor. The facility information 24 is associated with the sensor ID of at least one vibration sensor for each facility ID that identifies the facility.

　図３は、本実施形態の各装置のハードウェア構成を例示するブロック図である。各装置は、プロセッサ５０、メモリ５２、入出力インターフェイス（Ｉ／Ｆ）５４、周辺回路５６、バス５８を有する。周辺回路５６には、様々なモジュールが含まれる。処理装置は周辺回路５６を有さなくてもよい。 FIG. 3 is a block diagram illustrating the hardware configuration of each device of this embodiment. Each device has a processor 50, a memory 52, an input/output interface (I/F) 54, a peripheral circuit 56, and a bus 58. The peripheral circuit 56 includes various modules. The processing device may not have the peripheral circuit 56.

　バス５８は、プロセッサ５０、メモリ５２、周辺回路５６及び入出力インターフェイス５４が相互にデータを伝送するためのデータ伝送路である。プロセッサ５０は、例えばＣＰＵ（Central Processing Unit）、ＧＰＵ（Graphics Processing Unit）などの演算処理装置である。メモリ５２、例えばＲＡＭ（Random Access Memory）やＲＯＭ（Read Only Memory）などのメモリである。入出力インターフェイス５４は、入力装置、外部装置、外部サーバ、センサ等から情報を取得するためのインターフェイスや、出力装置、外部装置、外部サーバ等に情報を出力するためのインターフェイスなどを含む。入力装置は、例えばキーボード、マウス、マイク等である。出力装置は、例えばディスプレイ、スピーカ、プリンタ、メーラ等である。プロセッサ５０は、各モジュールに指令を出し、それらの演算結果をもとに演算を行うことができる。 The bus 58 is a data transmission path for the processor 50, the memory 52, the peripheral circuit 56, and the input/output interface 54 to mutually transmit data. The processor 50 is an arithmetic processing device such as a CPU (Central Processing Unit) and a GPU (Graphics Processing Unit). The memory 52 is a memory such as a RAM (Random Access Memory) or a ROM (Read Only Memory). The input/output interface 54 includes an interface for acquiring information from an input device, an external device, an external server, a sensor, etc., an interface for outputting information to an output device, an external device, an external server, etc. The input device is, for example, a keyboard, a mouse, a microphone, or the like. The output device is, for example, a display, a speaker, a printer, a mailer, or the like. The processor 50 can issue a command to each module and perform a calculation based on the calculation result.

　後述する図４の本実施形態の解析装置１００の各構成要素は、図３に示すコンピュータのハードウェアとソフトウェアの任意の組合せによって実現される。そして、その実現方法、装置にはいろいろな変形例があることは、当業者には理解されるところである。以下説明する各実施形態の解析装置を示す機能ブロック図は、ハードウェア単位の構成ではなく、論理的な機能単位のブロックを示している。 Each component of the analysis device 100 of this embodiment of FIG. 4 described later is realized by an arbitrary combination of the hardware and software of the computer shown in FIG. It will be understood by those skilled in the art that there are various modified examples of the realizing method and the apparatus. The functional block diagram showing the analysis device of each embodiment described below shows blocks of logical functional units, not of hardware units.

　プロセッサ５０が、プログラムをメモリ５２に読み出して実行することにより、図４の解析装置１００の各ユニットの各機能を実現することができる。 By the processor 50 reading the program into the memory 52 and executing the program, each function of each unit of the analysis device 100 in FIG. 4 can be realized.

　本実施形態のコンピュータプログラムは、解析装置１００を実現させるためのコンピュータ（図３のプロセッサ５０）に、生産設備１０に設けられた振動センサ１２の検出結果を画像化する手順、画像化したデータを機械学習処理の対象にして判別器を生成する手順、判別器を用いて生産設備１０の異常判定処理を行う手順、を実行させるように記述されている。 The computer program of the present embodiment causes a computer (processor 50 in FIG. 3) for realizing the analysis device 100 to display a procedure for imaging the detection result of the vibration sensor 12 provided in the production facility 10 and the imaged data. It is described that a procedure for generating a discriminator as a target of machine learning processing and a procedure for performing an abnormality determination processing of the production facility 10 using the discriminator are executed.

　本実施形態のコンピュータプログラムは、コンピュータで読み取り可能な記録媒体に記録されてもよい。記録媒体は特に限定されず、様々な形態のものが考えられる。また、プログラムは、記録媒体からコンピュータのメモリ５２（図３）にロードされてもよいし、ネットワークを通じてコンピュータにダウンロードされ、メモリ５２にロードされてもよい。 The computer program of this embodiment may be recorded in a computer-readable recording medium. The recording medium is not particularly limited, and various forms are conceivable. Further, the program may be loaded from the recording medium into the memory 52 (FIG. 3) of the computer, or may be downloaded to the computer through the network and loaded into the memory 52.

　コンピュータプログラムを記録する記録媒体は、非一時的な有形のコンピュータが使用可能な媒体を含み、その媒体に、コンピュータが読み取り可能なプログラムコードが埋め込まれる。コンピュータプログラムが、コンピュータ上で実行されたとき、コンピュータに、解析装置１００を実現する本実施形態の解析方法を実行させる。 A recording medium for recording a computer program includes a non-transitory tangible computer-usable medium, and a computer-readable program code is embedded in the medium. When the computer program is executed on the computer, it causes the computer to execute the analysis method of the present embodiment that realizes the analysis device 100.

　図４は、本実施形態の解析装置１００の論理的な構成を示す機能ブロック図である。解析装置１００は、画像処理部１０２と、生成部１０４と、解析部１０６と、を備える。
　画像処理部１０２は、生産設備１０に設けられた振動センサの検出結果を画像化する。生成部１０４は、画像化したデータを機械学習処理の対象にして判別器１１０を生成する。解析部１０６は、判別器１１０を用いて生産設備１０の状態解析処理を行う。 FIG. 4 is a functional block diagram showing a logical configuration of the analysis device 100 of this embodiment. The analysis device 100 includes an image processing unit 102, a generation unit 104, and an analysis unit 106.
The image processing unit 102 visualizes the detection result of the vibration sensor provided in the production facility 10. The generation unit 104 generates the discriminator 110 by using the imaged data as the target of machine learning processing. The analysis unit 106 uses the discriminator 110 to perform a state analysis process of the production facility 10.

　振動は、振幅、周波数、および位相の３つの要素からなり、測定データは、変位、速度、および加速度の３つのパラメータで示すことができる。振動センサの測定データは、複数のパラメータで示される複数の振動波形を含む時系列データであるため、そのまま機械学習にかけても特徴を捉えられず、適切な学習モデルを生成することができない。そこで、本実施形態では、画像処理部１０２により測定データを画像化する。図５は画像化前の測定データの時系列データを示す図である。図６は、画像処理部１０２により画像化した画像データを示す図である。 -Vibration consists of three elements: amplitude, frequency, and phase, and measurement data can be indicated by three parameters: displacement, velocity, and acceleration. Since the measurement data of the vibration sensor is time-series data including a plurality of vibration waveforms indicated by a plurality of parameters, the characteristics cannot be captured even by machine learning as it is, and an appropriate learning model cannot be generated. Therefore, in the present embodiment, the measurement data is imaged by the image processing unit 102. FIG. 5 is a diagram showing time-series data of measurement data before imaging. FIG. 6 is a diagram showing image data imaged by the image processing unit 102.

　画像処理部１０２は、生産設備１０のセンサ１２の検出結果（以下、振動データとも呼ぶ）を取得する。そして、画像処理部１０２は、取得した振動データをＦＦＴ処理してから周波数分割し、そして、得られた振動スペクトラムデータを画像化して画像データを得る。これらの処理により、データ容量を圧縮することができ、機械学習処理および判別処理を高速化することができる。 The image processing unit 102 acquires a detection result of the sensor 12 of the production facility 10 (hereinafter, also referred to as vibration data). Then, the image processing unit 102 performs FFT processing on the acquired vibration data, frequency-divides the acquired vibration data, and images the acquired vibration spectrum data to obtain image data. By these processes, the data capacity can be compressed, and the machine learning process and the discrimination process can be speeded up.

　実施形態において「取得」とは、自装置が他の装置や記憶媒体に格納されているデータまたは情報を取りに行くこと（能動的な取得）、および、自装置に他の装置から出力されるデータまたは情報を入力すること（受動的な取得）の少なくとも一方を含む。能動的な取得の例は、他の装置にリクエストまたは問い合わせしてその返信を受信すること、及び、他の装置や記憶媒体にアクセスして読み出すこと等がある。また、受動的な取得の例は、配信（または、送信、プッシュ通知等）される情報を受信すること等がある。さらに、「取得」とは、受信したデータまたは情報の中から選択して取得すること、または、配信されたデータまたは情報を選択して受信することであってもよい。 In the embodiment, “acquisition” means that the device itself acquires data or information stored in another device or a storage medium (active acquisition), and is output to the device itself from another device. At least one of inputting data or information (passive acquisition) is included. Examples of active acquisition include requesting or inquiring another device to receive the reply, and accessing and reading the other device or a storage medium. An example of passive acquisition is receiving information to be distributed (or transmitted, push notification, etc.). Further, “acquisition” may mean selecting and acquiring from received data or information, or selecting and receiving distributed data or information.

　判別器１１０は、生成部１０４により、画像化されたデータを機械学習処理の対象にして生成される。解析部１０６は、この判別器１１０を用いて、生産設備１０の状態解析処理を行う。 The discriminator 110 is generated by the generation unit 104 using the imaged data as the target of machine learning processing. The analysis unit 106 uses the discriminator 110 to perform a state analysis process of the production facility 10.

　機械学習は、例えば、ディープラーニング（深層学習）を用いることができるが、これに限定されない。判別器１１０が判別する生産設備１０の状態は、例えば、正常か、それ以外の状態である。以後、正常以外の状態は「Unknown」とも呼ぶ。本実施形態では、判別器１１０は異常状態の判別は行わない。判別器１１０の機械学習処理については後述する実施形態で詳細に説明する。 For machine learning, for example, deep learning can be used, but it is not limited to this. The state of the production facility 10 determined by the discriminator 110 is, for example, normal or any other state. Hereinafter, states other than normal are also referred to as “Unknown”. In this embodiment, the discriminator 110 does not discriminate an abnormal state. The machine learning process of the discriminator 110 will be described in detail in an embodiment described later.

　判別器１１０の判別結果はオペレータが参照できるように出力されてもよい。出力方法は、例えば、解析装置１００のディスプレイに表示してもよいし、解析装置１００からプリンタに印字出力されてもよいし、通信回線を介して他の装置（例えば、操作端末等）に送信されてもよい。 The discrimination result of the discriminator 110 may be output so that the operator can refer to it. The output method may be displayed on the display of the analysis device 100, may be printed out from the analysis device 100 on a printer, or may be transmitted to another device (for example, an operation terminal) via a communication line. May be done.

　このように構成された本実施形態の解析装置１００の動作について説明する。図７は、本実施形態の解析装置１００の動作の一例を示すフローチャートである。
　まず、画像処理部１０２は、生産設備１０に設けられた振動センサの検出結果を画像化する（ステップＳ１０１）。そして、生成部１０４は、画像化したデータを機械学習処理の対象にして判別器１１０を生成する（ステップＳ１０３）。そして、解析部１０６は、判別器１１０を用いて生産設備１０の状態解析処理を行う（ステップＳ１０５）。 The operation of the analyzing apparatus 100 of this embodiment configured as above will be described. FIG. 7 is a flowchart showing an example of the operation of the analysis device 100 of this embodiment.
First, the image processing unit 102 visualizes the detection result of the vibration sensor provided in the production facility 10 (step S101). Then, the generation unit 104 generates the discriminator 110 by using the imaged data as the target of the machine learning process (step S103). Then, the analysis unit 106 uses the discriminator 110 to perform a state analysis process of the production facility 10 (step S105).

　図８は、図７のステップＳ１０１の画像化処理の詳細フローの一例を示すフローチャートである。まず、画像処理部１０２は、センサ１２の測定データをＦＦＴ処理してから周波数分割し（ステップＳ１１３）、得られた振動スペクトラムデータを画像化して画像データを出力する（ステップＳ１１５）。ステップＳ１１５で得られた画像データは、図７のステップＳ１０３で解析部１０６により判別器１１０を用いた生産設備１０の状態解析処理が行われる。 FIG. 8 is a flowchart showing an example of a detailed flow of the imaging process of step S101 of FIG. First, the image processing unit 102 subjects the measurement data of the sensor 12 to FFT processing and frequency division (step S113), images the obtained vibration spectrum data, and outputs the image data (step S115). The image data obtained in step S115 is subjected to state analysis processing of the production facility 10 using the discriminator 110 by the analysis unit 106 in step S103 of FIG.

　以上説明したように、本実施形態において、画像処理部１０２によりセンサ１２の測定データが画像化された後、生成部１０４により画像化された画像データを機械学習処理の対象にして判別器１１０が生成され、そして、この判別器１１０を用いて解析部１０６により生産設備１０の状態解析処理が行われる。このように、本実施形態によれば、振動センサの検出結果を機械学習する際に、機械学習処理の対象となる振動波形データを画像化することで、データ量を圧縮できるので、処理負荷を低減するとともに、処理速度を高速化できる。 As described above, in the present embodiment, after the measurement data of the sensor 12 is imaged by the image processing unit 102, the discriminator 110 targets the image data imaged by the generation unit 104 as the target of the machine learning process. After being generated, the analysis unit 106 uses the discriminator 110 to perform a state analysis process of the production facility 10. As described above, according to this embodiment, when the detection result of the vibration sensor is machine-learned, the amount of data can be compressed by imaging the vibration waveform data that is the target of the machine-learning processing, so that the processing load is reduced. The processing speed can be increased as well as the processing speed is reduced.

（第２の実施の形態）
　図９は、本実施形態の解析装置１００の論理的な構成を示す機能ブロック図である。解析装置１００は、図４の解析装置１００と同様な画像処理部１０２と、生成部１０４と、解析部１０６と、を備えるとともに、さらに、異常判定部１２０を備える。 (Second embodiment)
FIG. 9 is a functional block diagram showing a logical configuration of the analysis device 100 of this embodiment. The analysis device 100 includes an image processing unit 102 similar to the analysis device 100 of FIG. 4, a generation unit 104, and an analysis unit 106, and further includes an abnormality determination unit 120.

　異常判定部１２０は、生産設備１０に設けられた振動センサの検出結果に基づいて当該生産設備１０の異常判定処理を第１の判別器１２４を用いて行う。異常判定部１２０は、第１の判別器１２４を用いた異常判定処理の結果を出力する。出力された結果は、設備監視システム１において生産設備１０の監視処理に使用されてもよい。 The abnormality determination unit 120 uses the first discriminator 124 to perform abnormality determination processing of the production facility 10 based on the detection result of the vibration sensor provided in the production facility 10. The abnormality determination unit 120 outputs the result of the abnormality determination process using the first discriminator 124. The output result may be used for the monitoring process of the production facility 10 in the facility monitoring system 1.

　また、本実施形態の画像処理部１０２は、画像化対象となる検出結果が、第１の判別器１２４により正常か異常かの判別ができなかった検出結果である点で上記実施形態の画像処理部１０２と相違する。生成部１０４は、このように閾値解析により正常か異常かの判別ができなかった測定データを機械学習処理するための第２の判別器１２６を生成する。 Further, the image processing unit 102 according to the present embodiment is that the detection result to be imaged is a detection result that the first discriminator 124 cannot determine whether the image processing unit 102 is normal or abnormal. Different from the part 102. The generation unit 104 generates the second discriminator 126 for performing the machine learning process on the measurement data which cannot be discriminated as normal or abnormal by the threshold analysis.

　振動センサで計測される振動は、複数の要因からなる複数の振動波形が含まれる。一般に、振動センサから検出される測定データは、高速フーリエ変換（ＦＦＴ： Fast Fourie Transform）処理を行うことにより周波数解析が行われる。測定データをＦＦＴ処理することにより、特徴点な周波数（ピーク）が検出され、検出されたピークレベルを閾値により正常か異常かを判別することで異常診断することが可能になる。 　The vibration measured by the vibration sensor contains multiple vibration waveforms consisting of multiple factors. Generally, frequency analysis is performed on measurement data detected by a vibration sensor by performing a Fast Fourier Transform (FFT) process. By performing FFT processing on the measurement data, the characteristic frequency (peak) is detected, and the abnormality can be diagnosed by determining whether the detected peak level is normal or abnormal by the threshold value.

　しかし、この閾値解析により正常か異常かの判別ができない場合がある。生成部１０４は、このように閾値解析により正常か異常かの判別できなかった測定データを機械学習処理の対象として第２の判別器１２６を生成する。この第２の判別器１２６は、図４の判別器１１０に相当する。 However, it may not be possible to distinguish between normal and abnormal by this threshold analysis. The generation unit 104 generates the second discriminator 126 by using the measurement data, which cannot be discriminated as normal or abnormal by the threshold analysis, as the target of the machine learning process. The second discriminator 126 corresponds to the discriminator 110 in FIG.

　「第１の判別器１２４で判別できない」場合とは、例えば、第１の判別モデル１２８に登録されている振動波形パターンと検出結果がマッチングしなかった場合、マッチング結果の尤度が所定値以上得られず、設備診断に必要とされる基準以上（例えば、検出率９０％以上）の信頼性が得られなかった場合、等である。 The case where “the first discriminator 124 cannot discriminate” means that, for example, when the vibration waveform pattern registered in the first discriminant model 128 and the detection result do not match, the likelihood of the matching result is equal to or more than a predetermined value. This is the case when the reliability is not obtained and the reliability higher than the standard required for equipment diagnosis (for example, the detection rate of 90% or more) is not obtained.

　解析部１０６は、第２の判別器１２６を用いて生産設備１０の状態解析処理を行う。解析部１０６は、第１の判別器１２４により正常か異常かの判別ができなかった検出結果を画像化したデータについて第２の判別器１２６を用いて生産設備１０の状態解析処理を行う。 The analysis unit 106 uses the second discriminator 126 to perform a state analysis process of the production facility 10. The analysis unit 106 uses the second discriminator 126 to analyze the state of the production facility 10 with respect to the data obtained by imaging the detection result in which the first discriminator 124 cannot discriminate between normal and abnormal.

　第１の判別器１２４は、第１の判別モデル１２８を用いて、正常か異常かを判別する。第１の判別モデル１２８は、例えば振動スペクトルのパターンマッチングや、閾値判定を用いたモデルである。 The first discriminator 124 uses the first discriminant model 128 to discriminate between normal and abnormal. The first discriminant model 128 is, for example, a model using pattern matching of vibration spectrum or threshold determination.

　一例として、第１の判別器１２４は、振動センサの測定データをＦＦＴ処理して得られた周波数分布から、特定周波数のピークレベル、最大および平均ピークレベルの比、Ｓ／Ｎ比（Signal-to-Noise ratio）、および特定周波数の範囲でのピークレベルの積算値の少なくともいずれか一つについてそれぞれ求め、求めた値に閾値を設定し、閾値の範囲内か否かに基づいて異常判定処理を行う。そして閾値範囲内の場合は、正常と判別し、閾値範囲外の場合は異常と判別する。 As an example, the first discriminator 124 uses the frequency distribution obtained by subjecting the measurement data of the vibration sensor to the FFT processing to determine the peak level of the specific frequency, the ratio of the maximum and average peak levels, and the S/N ratio (Signal-to -Noise ratio), and at least one of the integrated value of peak level in the range of the specific frequency, set the threshold value to the calculated value, and perform the abnormality determination process based on whether it is within the range of the threshold value. To do. If it is within the threshold range, it is determined to be normal, and if it is outside the threshold range, it is determined to be abnormal.

　第１の判別器１２４は複数の特定周波数を用いて上記した判定処理を行ってもよい。その場合は、第１の判別器１２４は特定周波数毎に判定結果を出力してもよい。また、第１の判別器１２４は、複数の特定周波数のうち１つでも閾値を外れた値があった場合に異常と判定してもよいし、複数の特定周波数のうちの所定数以上の特定周波数が閾値を外れた値があった場合に異常と判定してもよいし、複数の特定周波数のすべてが閾値を外れていた場合に異常と判定してもよいし、複数の特定周波数のすべてが閾値範囲内の場合に正常と判定してもよい。また、不具合事象毎に複数の特定周波数の値の組み合わせをモデル化して第１の判別モデル１２８に登録しておき、第１の判別器１２４は第１の判別モデル１２８とのパターンマッチングにより、不具合事象を判別してもよい。判別された不具合事象を、異常判定部１２０は、対応する不具合事象項目をオペレータに通知してもよい。 The first discriminator 124 may perform the above-described determination process using a plurality of specific frequencies. In that case, the first discriminator 124 may output the determination result for each specific frequency. Further, the first discriminator 124 may determine that there is an abnormality when even one of the plurality of specific frequencies is out of the threshold value, or may identify a predetermined number or more of the plurality of specific frequencies. It may be judged to be abnormal if there is a value outside the threshold value for the frequency, or may be judged to be abnormal if all of the multiple specific frequencies are outside the threshold value, or all of the multiple specific frequencies. May be determined to be normal when is within the threshold range. In addition, a combination of a plurality of values of specific frequencies is modeled for each fault event and registered in the first discriminant model 128, and the first discriminator 124 performs the pattern matching with the first discriminant model 128 to cause a defect. The event may be determined. For the determined malfunction event, the abnormality determination unit 120 may notify the operator of the corresponding malfunction event item.

　閾値は自動設定されてもよいし、オペレータによる手動設定で行われてもよい。自動設定の場合は、生成部１０４は、第１の判別モデル１２８に登録されている振動特性パターン（周波数分布）について、特定周波数のピークレベル、最大および平均ピークレベルの比、Ｓ／Ｎ比（Signal-to-Noise ratio）、および特定周波数の範囲でのピークレベルの積算値の少なくともいずれか一つについてそれぞれ求め、正常時と異常時の境界範囲を検出して閾値を設定してもよい。 Threshold may be set automatically or may be set manually by an operator. In the case of automatic setting, the generation unit 104, for the vibration characteristic pattern (frequency distribution) registered in the first discriminant model 128, the peak level of the specific frequency, the ratio of the maximum and average peak levels, and the S/N ratio ( Signal-to-Noise ratio) and the integrated value of the peak level in the range of the specific frequency, and the threshold value may be set by detecting the boundary range between the normal time and the abnormal time.

　手動設定の場合、生成部１０４は、第１の判別モデル１２８に登録されている振動特性パターン（周波数分布）について、特定周波数のピークレベル、最大および平均ピークレベルの比、Ｓ／Ｎ比（Signal-to-Noise ratio）、および特定周波数の範囲でのピークレベルの積算値の少なくともいずれか一つについてそれぞれ求めた結果をオペレータに提示して、第１の判別器１２４の閾値の設定をオペレータ操作によりそれぞれ受け付けてもよい。オペレータへの結果の提示方法は様々考えられるが、例として、解析装置１００のディスプレイに表示してもよいし、解析装置１００からプリンタに印字出力されてもよいし、通信回線を介して他の装置（例えば、操作端末等）に送信されてもよい。 In the case of manual setting, the generation unit 104, for the vibration characteristic pattern (frequency distribution) registered in the first discriminant model 128, the peak level of the specific frequency, the ratio of the maximum and average peak levels, and the S/N ratio (Signal). -to-Noise ratio) and at least one of the integrated value of the peak level in the range of the specific frequency are presented to the operator, and the operator sets the threshold value of the first discriminator 124. May be accepted respectively. There are various possible methods of presenting the result to the operator, but as an example, the result may be displayed on the display of the analysis device 100, may be printed out from the analysis device 100 on a printer, or may be output via a communication line. It may be transmitted to a device (for example, an operation terminal or the like).

　第２の判別器１２６は、生成部１０４により生成され、上記実施形態の生成部１０４で生成される判別器１１０に相当する。生成部１０４は、第２の判別器１２６により正常と判別されたデータのみを機械学習させ、正常状態を正規化して第２の判別モデル１３０を更新する。ここで、正規化とは、正常時の振動センサの測定データの画像化データのパターンをモデル化することである。第２の判別器１２６は、第２の判別モデル１３０の範囲から外れたデータを、例えば、「Unknown」として抽出する。つまり、この第２の判別器１２６を用いて解析部１０６により、第１の判別器１２４で判別できなかった検出結果について、さらに、第２の判別モデル１３０に基づいて、生産設備１０の状態が、正常か「Unknown」かが判別される。第２の判別器１２６は、生産設備１０の異常状態は判別しない、正常以外の状態は「Unknown」と判別する。 The second discriminator 126 corresponds to the discriminator 110 generated by the generation unit 104 and generated by the generation unit 104 in the above embodiment. The generator 104 machine-learns only the data discriminated as normal by the second discriminator 126, normalizes the normal state, and updates the second discriminant model 130. Here, the normalization is to model the pattern of the imaging data of the measurement data of the vibration sensor at the normal time. The second discriminator 126 extracts data outside the range of the second discriminant model 130, for example, as “Unknown”. That is, with respect to the detection result which cannot be discriminated by the first discriminator 124 by the analysis unit 106 using the second discriminator 126, the state of the production facility 10 is further determined based on the second discriminant model 130. , Normal or "Unknown" is determined. The second discriminator 126 does not discriminate an abnormal state of the production facility 10, and discriminates a state other than normal as “Unknown”.

　この判別器１１０により「Unknown」と判別されたデータは、オペレータにより参照されて、分析されてもよい。オペレータにより分析された結果は、検出結果の異常判定処理の閾値に反映されてもよい。「Unknown」と判別されたデータの利用方法については後述する実施形態で詳細に説明する。 The data discriminated as “Unknown” by the discriminator 110 may be referred to and analyzed by the operator. The result analyzed by the operator may be reflected in the threshold value of the abnormality determination process of the detection result. A method of using the data determined as “Unknown” will be described in detail in an embodiment described later.

　このように構成された本実施形態の解析装置１００の動作について説明する。図１０は、本実施形態の解析装置１００の動作の一例を示すフローチャートである。
　まず、異常判定部１２０は、生産設備１０に設けられた振動センサの検出結果に基づいて当該生産設備１０の異常判定処理を第１の判別器１２４を用いて行う（ステップＳ１２１）。そして、画像処理部１０２は、第１の判別器１２４により正常か異常かの判別ができなかった検出結果を（ステップＳ１２３のＮＯ）、画像化する（ステップＳ１２５）。判別できた場合は（ステップＳ１２３のＹＥＳ）、本処理を終了する。 The operation of the analyzing apparatus 100 of this embodiment configured as above will be described. FIG. 10 is a flowchart showing an example of the operation of the analysis device 100 of this embodiment.
First, the abnormality determination unit 120 uses the first discriminator 124 to perform abnormality determination processing of the production facility 10 based on the detection result of the vibration sensor provided in the production facility 10 (step S121). Then, the image processing unit 102 visualizes the detection result of which the first discriminator 124 could not determine whether it is normal or abnormal (NO in step S123) (step S125). If it can be determined (YES in step S123), this process ends.

　そして、生成部１０４は、画像化したデータを機械学習処理の対象にして第２の判別器１２６を生成する（ステップＳ１２７）。そして、解析部１０６は、第２の判別器１２６を用いて生産設備１０の状態解析処理を行う（ステップＳ１２９）。 Then, the generation unit 104 generates the second discriminator 126 by using the imaged data as the target of the machine learning process (step S127). Then, the analysis unit 106 uses the second discriminator 126 to perform a state analysis process of the production facility 10 (step S129).

　また、本実施形態のコンピュータプログラムは、解析装置１００を実現させるためのコンピュータ（図３のプロセッサ５０）に、生産設備１０に設けられた振動センサ１２の検出結果に基づいて当該生産設備１０の異常判定処理を第１の判別器１２４を用いて行う手順、第１の判別器１２４により正常か異常かの判別ができなかった検出結果を画像化する手順、画像化したデータを機械学習処理の対象として第２の判別器１２６を生成する手順、第２の判別器１２６を用いて生産設備１０の状態解析処理を行う手順、を実行させるように記述されている。 Further, the computer program of the present embodiment causes the computer (processor 50 in FIG. 3) for realizing the analysis apparatus 100 to detect an abnormality in the production facility 10 based on the detection result of the vibration sensor 12 provided in the production facility 10. A procedure for performing the determination process using the first discriminator 124, a procedure for imaging the detection result in which the first discriminator 124 cannot determine whether it is normal or abnormal, and the imaged data is the target of the machine learning process. It is described that the procedure for generating the second discriminator 126 and the procedure for performing the state analysis process of the production facility 10 by using the second discriminator 126 are executed.

　以上説明したように、本実施形態において、異常判定部１２０により第１の判別器１２４を用いて振動センサの検出結果に基づく異常判定処理で判別結果が得られなかった場合に、そのときの振動センサの検出結果を画像処理部１０２により画像化し、第２の判別器１２６により正常か「Unknown」かが判別される。このように、本実施形態によれば、第２の判別器１２６には、正常状態に対応する振動センサの検出結果のみを機械学習させ、正常状態について正規化するので、少量多品種や変種変動生産などで、製造条件が常に流動的が不具合事象の情報が積み上がりにくい場合であっても、判別の精度を向上できる。 As described above, in the present embodiment, when the abnormality determination unit 120 uses the first discriminator 124 to obtain no determination result in the abnormality determination process based on the detection result of the vibration sensor, the vibration at that time is detected. The detection result of the sensor is imaged by the image processing unit 102, and the second discriminator 126 discriminates whether it is normal or “Unknown”. As described above, according to the present embodiment, the second discriminator 126 performs machine learning only on the detection result of the vibration sensor corresponding to the normal state and normalizes the normal state. Even in the case where the manufacturing conditions are always in fluidity such as in production, and it is difficult to accumulate the information on the failure event, the accuracy of the determination can be improved.

　また、上記実施形態と同様に、振動センサの検出結果を機械学習する際に、機械学習処理の対象となる振動波形データを画像化することで、データ量を圧縮できるので、処理負荷を低減するとともに、処理速度を高速化できる。 Further, similarly to the above-described embodiment, when the detection result of the vibration sensor is machine-learned, the amount of data can be compressed by imaging the vibration waveform data that is the target of the machine-learning process, so that the processing load is reduced. At the same time, the processing speed can be increased.

（第３の実施の形態）
　図１１は、本実施形態の解析装置１００における異常判定処理の手順の一例を示すフローチャートである。本実施形態は、第１の判別器１２４を用いた生産設備１０の異常判定処理の結果を出力する構成を有する点以外は図９の解析装置１００と同様である。 (Third Embodiment)
FIG. 11 is a flowchart showing an example of a procedure of abnormality determination processing in the analysis device 100 of this embodiment. The present embodiment is the same as the analysis apparatus 100 in FIG. 9 except that it has a configuration for outputting the result of the abnormality determination processing of the production facility 10 using the first discriminator 124.

　まず、異常判定部１２０は、生産設備１０に設けられた振動センサから振動データを取得する（ステップＳ３０１）。そして、異常判定部１２０は、第１の判別器１２４を用いて取得した振動データの異常判定処理を実行する（ステップＳ３０３）。ここで、振動データはＦＦＴ処理された後、異常判定部１２０の第１の判別モデル１２８によって正常か異常を判別される（ステップＳ３０５）。 First, the abnormality determination unit 120 acquires vibration data from the vibration sensor provided in the production facility 10 (step S301). Then, the abnormality determination unit 120 executes the abnormality determination process of the vibration data acquired using the first discriminator 124 (step S303). Here, after the vibration data is subjected to FFT processing, the first discriminant model 128 of the abnormality determination unit 120 determines whether the vibration data is normal or abnormal (step S305).

　異常判定部１２０は、ＦＦＴ処理して得られた周波数分布から、特定周波数のピークレベル、最大および平均ピークレベルの比、Ｓ／Ｎ比（Signal-to-Noise ratio）、および特定周波数の範囲でのピークレベルの積算値の少なくともいずれか一つについてそれぞれ求め、閾値を用いてこれらの値が正常か異常かを判別する（ステップＳ３０５）。閾値の範囲内の場合（ステップＳ３０５のＮＯ）、第１の判別器１２４は、生産設備１０は正常であると判定し（ステップＳ３０７）、異常判定部１２０は、生産設備１０は正常であることを示す判定結果を出力する（ステップＳ３１１）。 From the frequency distribution obtained by the FFT processing, the abnormality determination unit 120 determines the peak level of the specific frequency, the ratio of the maximum and average peak levels, the S/N ratio (Signal-to-Noise ratio), and the range of the specific frequency. For each of at least any one of the integrated values of the peak levels, the threshold value is used to determine whether these values are normal or abnormal (step S305). If it is within the threshold range (NO in step S305), the first discriminator 124 determines that the production facility 10 is normal (step S307), and the abnormality determination unit 120 determines that the production facility 10 is normal. Is output (step S311).

　閾値の範囲外の場合（ステップＳ３０５のＹＥＳ）、第１の判別器１２４は、生産設備１０は異常状態であると判定し（ステップＳ３０９）、生産設備１０は異常状態であることを示す判定結果を出力する（ステップＳ３１１）。 If it is outside the threshold range (YES in step S305), the first discriminator 124 determines that the production facility 10 is in an abnormal state (step S309), and the determination result indicating that the production facility 10 is in an abnormal state. Is output (step S311).

　また、第１の判別器１２４による正常か異常かの判別ができなかった場合（ステップＳ３０５の判別不可）、異常判定部１２０は、画像処理部１０２に振動データを出力し（ステップＳ３１３）、図１０のステップＳ１２５に進む。 If the first discriminator 124 cannot determine whether it is normal or abnormal (determination in step S305 is not possible), the abnormality determination unit 120 outputs vibration data to the image processing unit 102 (step S313), and It progresses to step S125 of 10.

　本実施形態では、生産設備１０の異常判定処理に第１の判別器１２４を用い、第２の判別器１２６は異常判定処理には使用しない。また、第２の判別器１２６の判別結果は、第１の判別器１２４の第２の判別モデル１３０を更新するのに使用されるが、この構成については後述する実施形態で説明する。 In the present embodiment, the first discriminator 124 is used for the abnormality determination processing of the production facility 10, and the second discriminator 126 is not used for the abnormality determination processing. The discrimination result of the second discriminator 126 is used to update the second discriminant model 130 of the first discriminator 124, and this configuration will be described in an embodiment described later.

　以上説明したように、本実施形態において、異常判定部１２０により第１の判別器１２４を用いて生産設備１０の異常判定処理を行い、異常判定部１２０は第２の判別器１２６を生産設備１０の異常判定処理には使用しない。生成部１０４によって更新される第２の判別器１２６の判別結果を用いて第１の判別器１２４を更新することができる。この構成によれば、第２の判別器１２６の判別結果を反映した第１の判別器１２４を用いて異常判定処理を行うことができるので、判定結果の精度を向上できる。 As described above, in the present embodiment, the abnormality determination unit 120 performs the abnormality determination processing of the production facility 10 using the first discriminator 124, and the abnormality determination unit 120 causes the second discriminator 126 to operate the production facility 10. It is not used for the abnormality determination processing of. The first discriminator 124 can be updated using the discrimination result of the second discriminator 126 updated by the generation unit 104. According to this configuration, since the abnormality determination process can be performed using the first discriminator 124 that reflects the discrimination result of the second discriminator 126, the accuracy of the determination result can be improved.

（第４の実施の形態）
　図１２は、本実施形態の解析装置１００の画像処理部１０２の論理的な構成を示す機能ブロック図である。本実施形態の解析装置１００は、画像処理部１０２において、測定データからノイズを除去する処理を行ってから、画像化処理を行う構成を有する点以外は上記実施形態と同様である。本実施形態の構成は、他のいずれの実施形態の構成と組み合わせてもよい。 (Fourth Embodiment)
FIG. 12 is a functional block diagram showing a logical configuration of the image processing unit 102 of the analysis device 100 of this embodiment. The analysis apparatus 100 of the present embodiment is the same as the above-described embodiment except that the image processing unit 102 has a configuration of performing noise removal from measurement data and then performing imaging processing. The configuration of this embodiment may be combined with the configuration of any other embodiment.

　画像処理部１０２は、ノイズ除去部１１２と、画像化処理部１１４とを含む。ノイズ除去部１１２は、検出結果に対してノイズ除去処理を行う。画像化処理部１１４は、ノイズ除去部１１２によるノイズ除去処理が行われた後の検出結果を画像化する。ノイズ除去により測定データの振動波形が鮮明になる。 The image processing unit 102 includes a noise removal unit 112 and an imaging processing unit 114. The noise removal unit 112 performs noise removal processing on the detection result. The imaging processing unit 114 images the detection result after the noise removal processing by the noise removal unit 112 is performed. By removing noise, the vibration waveform of measurement data becomes clear.

　ノイズ除去処理は、例えば、複数配置された振動センサからの環境ノイズを予め計測して記憶しておき、ノイズデータを元に差分処理することを含む。ただしノイズ除去処理は他の方法で行われてもよい。 The noise removal processing includes, for example, measuring and storing environmental noise from a plurality of arranged vibration sensors in advance and performing difference processing based on the noise data. However, the noise removal process may be performed by another method.

　図１３は、本実施形態の解析装置１００の画像処理部１０２の動作の一例を示すフローチャートである。図１３のフローチャートは、上記実施形態で説明した図８のフローチャートのステップＳ１１３およびステップＳ１１５に加え、さらにステップＳ１１１を含む。 FIG. 13 is a flowchart showing an example of the operation of the image processing unit 102 of the analysis device 100 of this embodiment. The flowchart of FIG. 13 includes step S111 in addition to steps S113 and S115 of the flowchart of FIG. 8 described in the above embodiment.

　ステップＳ１１１では、ノイズ除去部１１２は、測定データのノイズ除去処理を行う。そして、ステップＳ１１１でノイズ除去処理されたデータを画像化処理部１１４がＦＦＴ処理してから周波数分割処理し（ステップＳ１１３）、得られた振動スペクトラムデータを画像化して画像データを出力する（ステップＳ１１５）。ステップＳ１１５で得られた画像データは、図７のステップＳ１０３で解析部１０６により判別器１１０を用いた生産設備１０の状態解析処理が行われる。 In step S111, the noise removal unit 112 performs noise removal processing on measurement data. Then, the image processing unit 114 performs FFT processing on the noise-removed data in step S111 and frequency division processing (step S113), and the obtained vibration spectrum data is imaged to output the image data (step S115). ). The image data obtained in step S115 is subjected to state analysis processing of the production facility 10 using the discriminator 110 by the analysis unit 106 in step S103 of FIG.

　以上説明したように、本実施形態において、ノイズ除去部１１２によりノイズ除去処理されたセンサ１２の測定データを、画像化処理部１１４により画像化する。この構成により本実施形態によれば、上記実施形態と同様に機械学習処理又は判別処理の高速化を図ることができるとともに、さらに、ノイズ除去により測定データの振動波形が鮮明になり、ＦＦＴ処理および画像化処理の精度が向上し、測定データの解析結果の精度および信頼性が向上する。 As described above, in this embodiment, the measurement data of the sensor 12 that has been subjected to the noise removal processing by the noise removal unit 112 is imaged by the imaging processing unit 114. With this configuration, according to the present embodiment, it is possible to speed up the machine learning processing or the discrimination processing as in the above-described embodiment, and further, the vibration waveform of the measurement data becomes clear by the noise removal, and the FFT processing and The accuracy of the imaging process is improved, and the accuracy and reliability of the measurement data analysis result are improved.

（第５の実施の形態）
　図１４は、本実施形態の解析装置１００の論理的な構成を示す機能ブロック図である。本実施形態の解析装置１００は、上記実施形態とは、生産設備１０の状態解析処理により正常でないと判別されたデータを抽出し、そのデータに基づき第１の判別器１２４の第１の第１の判別モデル１２８を更新する構成を有する点以外は上記実施形態と同様である。 (Fifth Embodiment)
FIG. 14 is a functional block diagram showing a logical configuration of the analysis device 100 of this embodiment. The analysis device 100 of the present embodiment is different from the above-described embodiment in that it extracts data that is determined to be abnormal by the state analysis process of the production facility 10, and based on the data, the first first discriminator 124 of the first discriminator 124. The configuration is the same as that of the above embodiment except that the discriminant model 128 is updated.

　解析装置１００は、図９の解析装置１００と同様な画像処理部１０２と、生成部１０４と、解析部１０６と、異常判定部１２０と、を備えるとともに、さらに抽出部１４０を備える。 The analysis device 100 includes an image processing unit 102 similar to the analysis device 100 of FIG. 9, a generation unit 104, an analysis unit 106, and an abnormality determination unit 120, and further includes an extraction unit 140.

　抽出部１４０は、第２の判別器１２６を用いた状態解析処理で正常でない（「Unknown」）と判定されたデータを抽出する。生成部１０４は、抽出されたデータに基づく補正情報を受け付け第１の判別器１２４の第１の判別モデル１２８を更新する。ここで、補正情報は、不具合事象と振動特性とを紐付けた情報を含む。 The extraction unit 140 extracts data that is determined to be abnormal (“Unknown”) in the state analysis processing using the second discriminator 126. The generator 104 receives the correction information based on the extracted data and updates the first discriminant model 128 of the first discriminator 124. Here, the correction information includes information in which the malfunction event and the vibration characteristic are associated with each other.

　抽出されるデータとは、「Unknown」と判別された画像化データに対応する、画像処理部１０２が画像化処理する前の、振動センサから受信した生の振動波形データと、当該データの時刻情報を含む。この「Unknown」と判別された振動データが示す振動は、未だ特定されていない生産設備１０の不具合事象により引き起こされている可能性がある。 The extracted data means the raw vibration waveform data received from the vibration sensor before the image processing by the image processing unit 102 and the time information of the data, which corresponds to the imaged data determined as “Unknown”. including. The vibration indicated by the vibration data determined to be “Unknown” may be caused by a failure event of the production facility 10 that has not been specified yet.

　そこで、抽出された振動データを時刻情報とともに、設備監視システム１が有している生産設備１０の稼働情報、状態情報、加工物の情報等を用いて、オペレータが人手で分析を行い、当該振動の原因となった不具合事象を特定する。 Therefore, the operator manually analyzes the extracted vibration data together with the time information by using the operation information, the status information, the information of the workpiece, etc. of the production equipment 10 that the equipment monitoring system 1 has, and the vibration concerned. Identify the failure event that caused the.

　そのため、解析装置１００は、抽出部１４０が抽出したデータを出力し、オペレータに提示する。オペレータへのデータの提示方法は様々考えられるが、例として、解析装置１００のディスプレイに表示してもよいし、解析装置１００からプリンタに印字出力されてもよいし、通信回線を介して他の装置（例えば、操作端末等）に送信されてもよい。 Therefore, the analysis device 100 outputs the data extracted by the extraction unit 140 and presents it to the operator. Although various methods of presenting data to the operator can be considered, as an example, the data may be displayed on the display of the analysis device 100, may be printed out from the analysis device 100 on a printer, or may be output via a communication line. It may be transmitted to a device (for example, an operation terminal or the like).

　オペレータが特定した不具合情報に、対応する振動の振動特性情報を紐付けた図１５の補正情報３０を、オペレータは操作画面等を用いて解析装置１００に入力する。生成部１０４は、入力された補正情報３０を受け付け、第１の判別器１２４の第１の判別モデル１２８を更新する。 The operator inputs the correction information 30 of FIG. 15 in which the vibration characteristic information of the corresponding vibration is associated with the defect information specified by the operator to the analysis device 100 using an operation screen or the like. The generation unit 104 receives the input correction information 30 and updates the first discriminant model 128 of the first discriminator 124.

　図１６は、解析部１０６による判別処理の詳細フローの一例を示すフローチャートである。解析部１０６は、図１３のステップＳ１１５で出力された画像データを第２の判別器１２６にかけ、生産設備１０の状態解析処理を行う（ステップＳ４０１）。第２の判別器１２６により正常と判別されたデータは（ステップＳ４０１の「正常」）、生成部１０４に受け渡され、正常データとして機械学習処理される（ステップＳ４０３）。一方、解析部１０６は、第２の判別器１２６により正規化の範囲から外れたデータを（ステップＳ４０１の「Unknown」）、抽出する（ステップＳ４０５）。 FIG. 16 is a flowchart showing an example of a detailed flow of the discrimination processing by the analysis unit 106. The analysis unit 106 applies the image data output in step S115 of FIG. 13 to the second discriminator 126 to perform a state analysis process of the production facility 10 (step S401). The data determined to be normal by the second discriminator 126 (“normal” in step S401) is passed to the generation unit 104, and machine learning processing is performed as normal data (step S403). On the other hand, the analysis unit 106 extracts the data outside the normalization range by the second discriminator 126 (“Unknown” in step S401) (step S405).

　なお、図１６のフローを用いて、図８のステップＳ１１５で出力された画像データについても同様に処理してよい。 Note that the image data output in step S115 of FIG. 8 may be similarly processed using the flow of FIG.

　図１７は、図１６のステップＳ４０５で抽出されたデータを用いた判別器の更新処理手順の一例を示すフローチャートである。
　まず、抽出部１４０は、ステップＳ４０５で抽出したデータ（振動データと時刻情報）を出力する（ステップＳ４１１）。ここでは、例えば、解析装置１００のディスプレイに表示する。そして、オペレータは、ステップＳ４１１で表示された振動データを時刻情報とともに、設備監視システム１が有している生産設備１０の稼働情報、状態情報、加工物の情報等を用いて分析を行い、当該振動の原因となった不具合事象を特定する。オペレータは、特定した不具合事象と、対応する振動特性情報とを紐付けた情報を作成し、解析装置１００の操作画面に従い、第１の判別器１２４の閾値の補正情報として入力する。 FIG. 17 is a flowchart showing an example of a discriminator update processing procedure using the data extracted in step S405 of FIG.
First, the extraction unit 140 outputs the data (vibration data and time information) extracted in step S405 (step S411). Here, for example, it is displayed on the display of the analysis device 100. Then, the operator analyzes the vibration data displayed in step S411 together with the time information using the operation information of the production facility 10, the state information, the information of the workpiece, etc., which the facility monitoring system 1 has, Identify the faulty event that caused the vibration. The operator creates information in which the identified malfunction event and the corresponding vibration characteristic information are associated with each other, and inputs the information as correction information for the threshold of the first discriminator 124 according to the operation screen of the analysis device 100.

　生成部１０４は、入力された補正情報を受け付け（ステップＳ４１３）、受け付けた補正情報を用いて第１の判別器１２４を更新する（ステップＳ４１５）。 The generation unit 104 receives the input correction information (step S413), and updates the first discriminator 124 using the received correction information (step S415).

　具体的には、受け付けた補正情報に含まれる、振動特性情報と不具合事象とを第１の判別モデル１２８に登録し、特定周波数のピークレベル、最大および平均ピークレベルの比、Ｓ／Ｎ比（Signal-to-Noise ratio）、および特定周波数の範囲でのピークレベルの積算値の少なくともいずれか一つについてそれぞれ求め、正常時と異常時の境界範囲を検出して閾値を設定する。あるいは、オペレータが、上記特定した不具合事象に対応する振動特性情報から求められた各値を元に閾値をそれぞれ設定して操作画面を用いて入力してもよい。 Specifically, the vibration characteristic information and the malfunction event included in the received correction information are registered in the first discrimination model 128, and the peak level of the specific frequency, the ratio of the maximum and average peak levels, and the S/N ratio ( Signal-to-Noise ratio) and the integrated value of the peak level in the range of the specific frequency, and calculate the boundary range between normal and abnormal conditions and set the threshold. Alternatively, the operator may set thresholds based on the respective values obtained from the vibration characteristic information corresponding to the identified malfunction event, and input the thresholds using the operation screen.

　以上説明したように、本実施形態において、抽出部１４０により第２の判別器１２６の「Unknown」データが抽出され、オペレータに提示され、生成部１０４によりオペレータにより分析された不具合事象と振動特性情報とを紐付けた補正情報が受け付けられ、補正情報に基づき第１の判別器１２４が更新される。このように、本実施形態によれば、第１の判別器１２４が判別できなかった測定データについて、さらに第２の判別器１２６により判別処理を行い、「Unknown」とされたデータを抽出してオペレータが解析し、その結果を第１の判別器１２４に反映するので、異常判定処理の精度を向上させることができる。 As described above, in the present embodiment, the extracting unit 140 extracts the “Unknown” data of the second discriminator 126, presents it to the operator, and analyzes the operator by the generating unit 104. The correction information associated with and is received, and the first discriminator 124 is updated based on the correction information. As described above, according to this embodiment, the measurement data that the first discriminator 124 cannot discriminate is further discriminated by the second discriminator 126, and the data that is “Unknown” is extracted. Since the operator analyzes and reflects the result in the first discriminator 124, the accuracy of the abnormality determination processing can be improved.

　また、第２の判別器１２６には、正常時の情報のみを機械学習させるので、少量多品種や変種変動生産などで、製造条件が常に流動的が不具合事象の情報が積み上がりにくい場合であっても、第１の判別モデル１２８を更新できるので生産設備１０の異常状態の判定精度を向上させることができる。 Further, since the second discriminator 126 machine-learns only the information in the normal state, it may be the case that the manufacturing condition is always fluid and the information on the defective event is difficult to be accumulated in a small-lot, large-variety product or variable production. However, since the first determination model 128 can be updated, the accuracy of determining the abnormal state of the production facility 10 can be improved.

（第５の実施の形態の変形態様）
　図１４の解析装置１００は、図９の解析装置１００の構成に抽出部１４０を設けた構成としていた。その変形態様として、図４の解析装置１００において、抽出部１４０を設けた構成としてもよい。 (Modification of the fifth embodiment)
The analysis apparatus 100 of FIG. 14 has a configuration in which the extraction unit 140 is provided in the configuration of the analysis apparatus 100 of FIG. As a modification thereof, the analysis device 100 of FIG. 4 may be provided with the extraction unit 140.

　解析装置１００は、異常判定部１２０と、抽出部１４０と、さらに備える。異常判定部１２０は、検出結果が示す振動センサの振動の特徴解析処理を行い、閾値を用いて生産設備１０の異常判定処理を行う。抽出部１４０は、状態解析処理で生産設備１０が正常状態でないと判定されたデータを抽出する。生成部１０４は、抽出されたデータに基づく補正情報を受け付け、閾値を更新する。補正情報は、不具合事象と振動特性とを紐付けた情報を含む。 The analysis device 100 further includes an abnormality determination unit 120 and an extraction unit 140. The abnormality determination unit 120 performs the characteristic analysis process of the vibration of the vibration sensor indicated by the detection result, and performs the abnormality determination process of the production facility 10 using the threshold value. The extraction unit 140 extracts data that is determined by the state analysis process that the production facility 10 is not in a normal state. The generation unit 104 receives the correction information based on the extracted data and updates the threshold value. The correction information includes information in which the fault event and the vibration characteristic are associated with each other.

　以上、図面を参照して本発明の実施形態について述べたが、これらは本発明の例示であり、上記以外の様々な構成を採用することもできる。
　たとえば、図１８は、図１６の状態解析処理のステップＳ４０１で正常と判別された場合の処理手順の一例を示すフローチャートである。この実施形態は、第２の判別器１２６により正常と判別された検出結果を第２の判別器１２６の教師データとする構成を有する点以外は上記実施形態と同様である。 Although the embodiments of the present invention have been described above with reference to the drawings, these are merely examples of the present invention, and various configurations other than the above may be adopted.
For example, FIG. 18 is a flowchart showing an example of a processing procedure when it is determined as normal in step S401 of the state analysis processing of FIG. This embodiment is the same as the above embodiment except that the detection result determined by the second discriminator 126 is normal is used as the teacher data of the second discriminator 126.

　生成部１０４は、第２の判別器１２６を用いた状態解析処理により、正常と判定された検出結果を（図１６のステップＳ４０１の正常）、生産設備１０の正常状態の教師データとして第２の判別器１２６の第２の判別モデル１３０を更新する（ステップＳ５０１）。 The generation unit 104 uses the detection result determined to be normal by the state analysis processing using the second discriminator 126 (normal in step S401 of FIG. 16) as the second teacher data in the normal state of the production facility 10. The second discriminant model 130 of the discriminator 126 is updated (step S501).

　この構成によれば、第２の判別器１２６を用いた状態解析処理により、正常と判定された検出結果を生産設備１０の正常状態の教師データとして第２の判別モデル１３０が生成部１０４により更新される。このように、第１の判別器１２４による異常判定処理において判別できなかった測定データから正常状態の教師データを生成し、第２の判別モデル１３０を更新することができ、第２の判別器１２６の判別精度を向上することができる。 According to this configuration, the second discriminant model 130 is updated by the generation unit 104 with the detection result determined as normal by the state analysis processing using the second discriminator 126 as the teacher data of the normal state of the production facility 10. To be done. In this way, the teacher data in the normal state can be generated from the measurement data that could not be discriminated in the abnormality discriminating process by the first discriminator 124, and the second discriminant model 130 can be updated. It is possible to improve the determination accuracy of.

　さらに、上記実施形態の第１の判別器１２４と、第２の判別器１２６に加え、解析装置１００は、第３の判別器（不図示）を備えてもよい。図１９は、解析装置１００が第３の判別器を用いて不具合モデルを構築し、不具合事象を識別する処理手順の一例を示すフローチャートである。第３の判別器は、第１の判別器１２４で正常と判別された測定データを取得する（ステップＳ６０１）。さらに、第３の判別器は、第２の判別器１２６でUnknownと判別された測定データを取得する（ステップＳ６０３）。そして、第３の判別器は、これらのデータを機械学習し、不具合事象を分類したモデルを構築する（ステップＳ６０５）。なお、第３の判別器が機械学習する各測定データについても、上記実施形態で説明したノイズ除去処理および画像化処理を行ってよい。 Further, in addition to the first discriminator 124 and the second discriminator 126 of the above embodiment, the analysis device 100 may include a third discriminator (not shown). FIG. 19 is a flowchart showing an example of a processing procedure in which the analysis device 100 constructs a defect model using the third discriminator and identifies a defect event. The third discriminator acquires the measurement data discriminated as normal by the first discriminator 124 (step S601). Furthermore, the third discriminator acquires the measurement data discriminated as Unknown by the second discriminator 126 (step S603). Then, the third discriminator machine-learns these data and constructs a model in which defective events are classified (step S605). The noise removal processing and the imaging processing described in the above embodiment may be performed on each measurement data machine-learned by the third discriminator.

　さらに、第３の判別器を用いて、生産設備１０のセンサ１２の測定データが正常か異常かを判定し、さらに、異常の場合に不具合事象を識別する（ステップＳ６０７）。 Further, using the third discriminator, it is determined whether the measurement data of the sensor 12 of the production facility 10 is normal or abnormal, and if it is abnormal, the defective event is identified (step S607).

　また、複数のセンサ１２の各々の位置情報を設備情報２４に記憶しておき、振動データと位置情報の関係をさらに、機械学習させ、分類モデル２０２、第１の判別モデル１２８、および第２の判別モデル１３０の少なくともいずれか一つに反映させてもよい。 In addition, the position information of each of the plurality of sensors 12 is stored in the facility information 24, and the relationship between the vibration data and the position information is further machine-learned, and the classification model 202, the first discrimination model 128, and the second It may be reflected in at least one of the discrimination models 130.

　さらに、複数のセンサ１２の測定データの測定条件（設備種別、環境（温度、湿度）等）等の情報を設備情報２４に記憶しておいてもよい。この測定条件、生産設備１０の稼働情報に含まれる稼働状態、稼働条件等から近い条件の測定データをグルーピングして、グループ毎に測定データを機械学習させ、分類モデル２０２、第１の判別モデル１２８、および第２の判別モデル１３０の少なくともいずれか一つに反映させてもよい。 Furthermore, information such as measurement conditions (equipment type, environment (temperature, humidity), etc.) of measurement data of a plurality of sensors 12 may be stored in the equipment information 24. The measurement data is grouped with the measurement data of conditions close to the measurement condition, the operation condition included in the operation information of the production facility 10, the operation condition, etc., and the measurement data is machine-learned for each group, and the classification model 202 and the first discrimination model 128 are used. , And the second discriminant model 130 may be reflected.

（実施例１）
　図２０は、実施例１の解析装置を説明するためのフロー図である。
　まず、異常判定部１２０は、生産設備１０のセンサ１２から振動データが入力されると、第１の判別器１２４においてＦＦＴ処理を行う（ステップＳ１１）。このとき第１の判別モデル１２８を用いてパターンマッチング処理により振動特性を特定する。そして、第１の判別器１２４は振動特性が閾値の範囲内の場合は正常と判定し、閾値２１０の範囲を外れた場合は異常と判定する。異常判定部１２０はこの結果を設備監視システム１に生産設備１０の異常判定結果として出力する（不図示）。 (Example 1)
FIG. 20 is a flow chart for explaining the analyzing apparatus of the first embodiment.
First, when the vibration data is input from the sensor 12 of the production facility 10, the abnormality determination unit 120 performs the FFT processing in the first discriminator 124 (step S11). At this time, the vibration characteristic is specified by the pattern matching process using the first discrimination model 128. Then, the first discriminator 124 determines that the vibration characteristics are normal when the vibration characteristics are within the threshold range, and determines that the vibration characteristics are abnormal when the vibration characteristics are outside the threshold range. The abnormality determination unit 120 outputs this result to the equipment monitoring system 1 as an abnormality determination result of the production equipment 10 (not shown).

　さらに、異常判定部１２０は、ステップＳ１１で第１の判別器１２４により正常か否かの判別ができなかったデータを抽出し（ステップＳ１３）、第２の判別器１２６の機械学習処理の対象として解析部１０６に受け渡す。 Further, the abnormality determination unit 120 extracts the data for which the first discriminator 124 could not discriminate whether it is normal in step S11 (step S13), and sets it as the target of the machine learning process of the second discriminator 126. It is passed to the analysis unit 106.

　解析部１０６は、第１の判別器１２４で判別できなかったセンサ１２の測定データについてノイズ除去処理を行い（ステップＳ１５）、周波数分析して画像化する（ステップＳ１７）。解析部１０６は、画像化されたデータを第２の判別器１２６を用いて判別し（ステップＳ１９）、生産設備１０の状態が正常か否かを判別する（ステップＳ２１）。正常と判別された場合（ステップＳ２１のＹＥＳ）、生成部１０４は、当該正常と判別された測定データを機械学習し第２の判別器１２６の第２の判別モデル１３０を更新する（ステップＳ２３）。 The analysis unit 106 performs noise removal processing on the measurement data of the sensor 12 that cannot be discriminated by the first discriminator 124 (step S15), frequency-analyzes it, and forms an image (step S17). The analysis unit 106 determines the imaged data using the second discriminator 126 (step S19), and determines whether the state of the production facility 10 is normal (step S21). When it is determined to be normal (YES in step S21), the generation unit 104 machine-learns the measurement data determined to be normal, and updates the second discriminant model 130 of the second discriminator 126 (step S23). ..

　一方、正常と判別されなかった（「Unknown」）場合（ステップＳ２１のＮＯ）、抽出部１４０は、当該Unknownとなった測定データを抽出して出力する（ステップＳ３１）。オペレータは、この抽出されたUnknownなデータを参照し、生産設備１０の稼働情報等とともに分析して、不具合事象を特定する。そして、不具合事象と振動特性とを紐付けた補正情報を解析装置１００に入力する（ステップＳ３３）。 On the other hand, when it is not determined to be normal (“Unknown”) (NO in step S21), the extraction unit 140 extracts and outputs the unknown measurement data (step S31). The operator refers to the extracted unknown data and analyzes it together with the operation information of the production facility 10 and the like, and identifies the defective phenomenon. Then, the correction information in which the fault event and the vibration characteristic are associated with each other is input to the analysis device 100 (step S33).

　生成部１０４は、入力された補正情報を受け付け、受け付けた補正情報を元に第１の判別モデル１２８と閾値を更新する（ステップＳ３５）。 The generation unit 104 receives the input correction information, and updates the first discrimination model 128 and the threshold value based on the received correction information (step S35).

　このようにして、解析装置１００により、第１の判別器１２４で異常判定できなかった測定データを、第２の判別器１２６により機械学習することで、正常でないUnknownなデータを抽出し、生産設備１０の稼働情報とともにオペレータが分析することで不具合事象を特定し、振動特性と不具合事象を紐付けた補正情報として解析装置１００に入力することで、補正情報を元に第１の判別器１２４の第１の判別モデル１２８と閾値を更新することができる。 In this way, the analyzer 100 causes the second discriminator 126 to machine-learn the measurement data that could not be abnormally determined by the first discriminator 124, thereby extracting abnormal data that is not normal and producing the production facility. The operator analyzes the operation information together with the operation information of 10 to identify a defective event, and inputs the corrected information to the analysis apparatus 100 as the correction information in which the vibration characteristic and the defective event are associated with each other. The first discriminant model 128 and the threshold can be updated.

（実施例２）
　図２１は、実施例２の解析装置を説明するためのフロー図である。
　この実施例の解析装置１００は、第１の判別器１２４と、第２の判別器１２６と、に加え、さらに、第３の判別器２００を備える。
　第３の判別器２００は、第１の判別器１２４により正常と判別された測定データと、第２の判別器１２６によりUnknownと判別された測定データとを用いて機械学習する（ステップＳ４１）。第２の判別器１２６は、機械学習により、正常と異常の分類モデル２０２を構築する。分類モデル２０２は、不具合事象をクラス分けする。 (Example 2)
FIG. 21 is a flow chart for explaining the analyzing apparatus of the second embodiment.
The analysis apparatus 100 of this embodiment includes a third discriminator 200 in addition to the first discriminator 124 and the second discriminator 126.
The third discriminator 200 performs machine learning using the measurement data discriminated as normal by the first discriminator 124 and the measurement data discriminated as Unknown by the second discriminator 126 (step S41). The second discriminator 126 constructs a normal and abnormal classification model 202 by machine learning. The classification model 202 classifies defective events into classes.

　第３の判別器２００は、この分類モデル２０２を用いて、測定データが正常か異常の判別を行うとともに、さらに、不具合事象を識別して特定することができる。 Using the classification model 202, the third discriminator 200 can discriminate whether the measurement data is normal or abnormal, and can further discriminate and specify the faulty event.

　以上、実施形態および実施例を参照して本願発明を説明したが、本願発明は上記実施形態および実施例に限定されるものではない。本願発明の構成や詳細には、本願発明のスコープ内で当業者が理解し得る様々な変更をすることができる。
　なお、本発明において利用者に関する情報を取得、利用する場合は、これを適法に行うものとする。 Although the present invention has been described with reference to the exemplary embodiments and examples, the present invention is not limited to the above-described exemplary embodiments and examples. Various modifications that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the present invention.
In addition, in the present invention, when the information about the user is acquired and used, this is legally performed.

　上記の実施形態の一部または全部は、以下の付記のようにも記載されうるが、以下に限られない。
１．　生産設備に設けられた振動センサの検出結果を画像化する画像処理手段と、
　前記画像化したデータを機械学習処理の対象にして判別器を生成する生成手段と、
　前記判別器を用いて前記生産設備の状態解析処理を行う解析手段と、を備える解析装置。
２．　１．に記載の解析装置において、
　前記検出結果が示す前記振動センサの振動の特徴解析処理を行い、閾値を用いて前記生産設備の異常判定処理を行う判定手段と、
　前記状態解析処理で前記生産設備が正常状態でないと判定されたデータを抽出する抽出手段と、をさらに備え、
　前記生成手段は、抽出された前記データに基づく補正情報を受け付け、前記閾値を更新し、
　前記補正情報は、不具合事象と振動特性とを紐付けた情報を含む、解析装置。
３．　生産設備に設けられた振動センサの検出結果に基づいて当該生産設備の異常判定処理を第１の判別器を用いて行う判定手段と、
　前記第１の判別器により正常か異常かの判別ができなかった前記検出結果を画像化する画像処理手段と、
　前記画像化したデータを機械学習処理の対象として第２の判別器を生成する生成手段と、
　前記第２の判別器を用いて前記生産設備の状態解析処理を行う解析手段と、
を備える、
解析装置。
４．　３．に記載の解析装置において、
　前記状態解析処理で前記生産設備が正常状態でないと判定されたデータを抽出する抽出手段をさらに備え、
　前記生成手段は、抽出された前記データに基づく補正情報を受け付け前記第１の判別器を更新し、
　前記補正情報は、不具合事象と振動特性とを紐付けた情報を含む、解析装置。
５．　１．から４．のいずれか一つに記載の解析装置において、
　前記生成手段は、前記状態解析処理により正常と判定された前記データを前記機械学習処理の教師データとする、解析装置。
６．　１．から５．のいずれか一つに記載の解析装置において、
　前記検出結果に対してノイズ除去処理を行う処理手段をさらに備え、
　前記画像処理手段は、前記処理手段による前記ノイズ除去処理が行われた後の前記検出結果を画像化する、解析装置。
７．　１．から６．のいずれか一つに記載の解析装置において、
　前記生産設備は、ベルトコンベアであり、
　前記振動センサは、前記ベルトコンベアに設けられた複数の振動センサである、解析装置。 The whole or part of the exemplary embodiments disclosed above can be described as, but not limited to, the following supplementary notes.
1. Image processing means for imaging the detection result of the vibration sensor provided in the production facility,
Generating means for generating a discriminator by subjecting the imaged data to machine learning processing;
And an analysis unit that performs a state analysis process of the production facility using the discriminator.
2. 1. In the analyzer described in
A determination unit that performs a characteristic analysis process of the vibration of the vibration sensor indicated by the detection result, and performs an abnormality determination process of the production facility using a threshold value,
Further comprising an extracting unit for extracting the data determined in the state analysis process that the production facility is not in a normal state,
The generation means receives correction information based on the extracted data, updates the threshold value,
The said correction information is an analysis apparatus containing the information which linked|related the malfunction event and the vibration characteristic.
3. Determination means for performing abnormality determination processing of the production facility using the first discriminator based on the detection result of the vibration sensor provided in the production facility;
Image processing means for imaging the detection result, which cannot be discriminated as normal or abnormal by the first discriminator;
Generating means for generating a second discriminator using the imaged data as a target of machine learning processing;
Analysis means for performing a state analysis process of the production facility using the second discriminator;
With
Analyzer.
4. 3. In the analyzer described in
The state analysis process further comprises an extracting unit for extracting data determined to be in a non-normal state of the production facility,
The generation means receives correction information based on the extracted data, updates the first discriminator,
The said correction information is an analysis apparatus containing the information which linked|related the malfunction event and the vibration characteristic.
5. 1. To 4. In the analysis device according to any one of
The analysis device, wherein the generation unit uses the data determined to be normal by the state analysis process as teacher data for the machine learning process.
6. 1. To 5. In the analysis device according to any one of
Further comprising processing means for performing noise removal processing on the detection result,
The analysis device, wherein the image processing means images the detection result after the noise removal processing is performed by the processing means.
7. 1. To 6. In the analysis device according to any one of
The production facility is a belt conveyor,
The said vibration sensor is an analysis apparatus which is a some vibration sensor provided in the said belt conveyor.

８．　解析装置が、
　生産設備に設けられた振動センサの検出結果を画像化し、
　前記画像化したデータを機械学習処理の対象にして判別器を生成し、
　前記判別器を用いて前記生産設備の状態解析処理を行う、
　解析方法。
９．　８．に記載の解析方法において、
　前記解析装置が、さらに、
　前記検出結果が示す前記振動センサの振動の特徴解析処理を行い、閾値を用いて前記生産設備の異常判定処理を行い、
　前記状態解析処理で前記生産設備が正常状態でないと判定されたデータを抽出し、
　抽出された前記データに基づく補正情報を受け付け、前記閾値を更新し、
　前記補正情報は、不具合事象と振動特性とを紐付けた情報を含む、解析方法。
１０．　解析装置が、
　生産設備に設けられた振動センサの検出結果に基づいて当該生産設備の異常判定処理を第１の判別器を用いて行い、
　前記第１の判別器により正常か異常かの判別ができなかった前記検出結果を画像化し、
　前記画像化したデータを機械学習処理の対象として第２の判別器を生成し、
　前記第２の判別器を用いて前記生産設備の状態解析処理を行う、
　解析方法。
１１．　１０．に記載の解析方法において、
　前記解析装置が、さらに、
　前記状態解析処理で前記生産設備が正常状態でないと判定されたデータを抽出し、
　抽出された前記データに基づく補正情報を受け付け前記第１の判別器を更新し、
　前記補正情報は、不具合事象と振動特性とを紐付けた情報を含む、解析方法。
１２．　８．から１１．のいずれか一つに記載の解析方法において、
　前記解析装置が、さらに、
　前記状態解析処理により正常と判定された前記データを前記機械学習処理の教師データとする、解析方法。
１３．　８．から１２．のいずれか一つに記載の解析方法において、
　前記解析装置が、さらに、
　前記検出結果に対してノイズ除去処理を行い、
　前記ノイズ除去処理が行われた後の前記検出結果を画像化する、解析方法。
１４．　８．から１３．のいずれか一つに記載の解析方法において、
　前記生産設備は、ベルトコンベアであり、
　前記振動センサは、前記ベルトコンベアに設けられた複数の振動センサである、解析方法。 8. The analysis device
Visualize the detection result of the vibration sensor installed in the production facility,
Generate a discriminator with the imaged data as the target of machine learning processing,
Performing a state analysis process of the production facility using the discriminator,
analysis method.
9. 8. In the analysis method described in
The analysis device further comprises
Performing a characteristic analysis process of the vibration of the vibration sensor indicated by the detection result, performing an abnormality determination process of the production facility using a threshold,
Extracting the data determined that the production equipment is not in a normal state in the state analysis process,
Accept correction information based on the extracted data, update the threshold,
The said correction information is an analysis method containing the information which linked|related the malfunction event and the vibration characteristic.
10. The analysis device
Based on the detection result of the vibration sensor provided in the production equipment, the abnormality determination processing of the production equipment is performed using the first discriminator,
The detection result, which could not be discriminated as normal or abnormal by the first discriminator, is imaged,
A second discriminator is generated by using the imaged data as a target of machine learning processing,
Performing a state analysis process of the production facility using the second discriminator,
analysis method.
11. 10. In the analysis method described in
The analysis device further comprises
Extracting the data determined that the production equipment is not in a normal state in the state analysis process,
Accepting correction information based on the extracted data, updating the first discriminator,
The said correction information is an analysis method containing the information which linked|related the malfunction event and the vibration characteristic.
12. 8. To 11. In the analysis method described in any one of
The analysis device further comprises
An analysis method, wherein the data determined to be normal by the state analysis process is used as teacher data for the machine learning process.
13. 8. To 12. In the analysis method described in any one of
The analysis device further comprises
Noise removal processing is performed on the detection result,
An analysis method of imaging the detection result after the noise removal processing is performed.
14. 8. To 13. In the analysis method described in any one of
The production facility is a belt conveyor,
An analysis method, wherein the vibration sensor is a plurality of vibration sensors provided on the belt conveyor.

１５．　コンピュータに、
　生産設備に設けられた振動センサの検出結果を画像化する手順、
　前記画像化したデータを機械学習処理の対象にして判別器を生成する手順、
　前記判別器を用いて前記生産設備の異常判定処理を行う手順、を実行させるためのプログラム。
１６．　１５．に記載のプログラムにおいて、
　前記検出結果が示す前記振動センサの振動の特徴解析処理を行い、閾値を用いて前記生産設備の異常判定処理を行う手順、
　前記状態解析処理で前記生産設備が正常状態でないと判定されたデータを抽出する手順、
　前記生成する手順において、抽出された前記データに基づく補正情報を受け付け、前記閾値を更新する手順、をさらにコンピュータに実行させ、
　前記補正情報は、不具合事象と振動特性とを紐付けた情報を含む、プログラム。
１７．　コンピュータに、
　生産設備に設けられた振動センサの検出結果に基づいて当該生産設備の異常判定処理を第１の判別器を用いて行う手順、
　前記第１の判別器により正常か異常かの判別ができなかった前記検出結果を画像化する手順、
　前記画像化したデータを機械学習処理の対象として第２の判別器を生成する手順、
　前記第２の判別器を用いて前記生産設備の状態解析処理を行う手順、を実行させるためのプログラム。
１８．　１７．に記載のプログラムにおいて、
　前記状態解析処理で前記生産設備が正常状態でないと判定されたデータを抽出する手順、
　前記生成する手順において、抽出された前記データに基づく補正情報を受け付け前記第１の判別器を更新する手順、をさらにコンピュータに実行させ、
　前記補正情報は、不具合事象と振動特性とを紐付けた情報を含む、プログラム。
１９．　１５．から１８．のいずれか一つに記載のプログラムにおいて、
　前記生成する手順において、前記状態解析処理により正常と判定された前記データを前記機械学習処理の教師データとする手順をさらにコンピュータに実行させるためのプログラム。
２０．　１５．から１９．のいずれか一つに記載のプログラムにおいて、
　前記検出結果に対してノイズ除去処理を行う手順、
　前記画像化する手順において、前記ノイズ除去処理が行われた後の前記検出結果を画像化する手順、をさらにコンピュータに実行させるためのプログラム。
２１．　１５．から２０．のいずれか一つに記載のプログラムにおいて、
　前記生産設備は、ベルトコンベアであり、
　前記振動センサは、前記ベルトコンベアに設けられた複数の振動センサである、プログラム。 15. On the computer,
Procedure for imaging the detection result of the vibration sensor installed in the production facility,
A procedure for generating a discriminator by subjecting the imaged data to machine learning processing,
A program for executing a procedure of performing abnormality determination processing of the production facility using the discriminator.
16. 15. In the program described in
Performing a characteristic analysis process of the vibration of the vibration sensor indicated by the detection result, a procedure for performing an abnormality determination process of the production facility using a threshold,
A procedure for extracting data that is determined in the state analysis process that the production facility is not in a normal state,
In the step of generating, the correction information based on the extracted data is accepted, and the procedure of updating the threshold value is further executed by a computer,
The said correction information is a program containing the information which linked|related the malfunction event and the vibration characteristic.
17. On the computer,
A procedure for performing abnormality determination processing of the production facility using the first discriminator based on the detection result of the vibration sensor provided in the production facility,
A procedure for imaging the detection result, which cannot be discriminated as normal or abnormal by the first discriminator,
A procedure for generating a second discriminator using the imaged data as a target of machine learning processing;
A program for executing a procedure for performing a state analysis process of the production facility using the second discriminator.
18. 17. In the program described in
A procedure for extracting data that is determined in the state analysis process that the production facility is not in a normal state,
In the step of generating, further causing a computer to execute a step of receiving correction information based on the extracted data and updating the first discriminator,
The said correction information is a program containing the information which linked|related the malfunction event and the vibration characteristic.
19. 15. To 18. In the program described in any one of,
A program for causing a computer to further execute a procedure of using the data determined to be normal by the state analysis processing as teacher data of the machine learning processing in the generation procedure.
20. 15. To 19. In the program described in any one of,
A procedure for performing noise removal processing on the detection result,
A program for causing a computer to further execute the step of imaging the detection result after the noise removal processing is performed in the step of imaging.
21. 15. To 20. In the program described in any one of,
The production facility is a belt conveyor,
A program in which the vibration sensor is a plurality of vibration sensors provided on the belt conveyor.

　この出願は、２０１９年２月５日に出願された日本出願特願２０１９－０１９０６８号を基礎とする優先権を主張し、その開示の全てをここに取り込む。 This application claims priority based on Japanese Patent Application No. 2019-019068 filed on February 5, 2019, and incorporates all the disclosure thereof.

Claims (21)

　生産設備に設けられた振動センサの検出結果を画像化する画像処理手段と、
　前記画像化したデータを機械学習処理の対象にして判別器を生成する生成手段と、
　前記判別器を用いて前記生産設備の状態解析処理を行う解析手段と、を備える解析装置。 Image processing means for imaging the detection result of the vibration sensor provided in the production facility,
Generating means for generating a discriminator by subjecting the imaged data to machine learning processing;
And an analysis unit that performs a state analysis process of the production facility using the discriminator. 　請求項１に記載の解析装置において、
　前記検出結果が示す前記振動センサの振動の特徴解析処理を行い、閾値を用いて前記生産設備の異常判定処理を行う判定手段と、
　前記状態解析処理で前記生産設備が正常状態でないと判定されたデータを抽出する抽出手段と、をさらに備え、
　前記生成手段は、抽出された前記データに基づく補正情報を受け付け、前記閾値を更新し、
　前記補正情報は、不具合事象と振動特性とを紐付けた情報を含む、解析装置。 The analysis device according to claim 1,
A determination unit that performs a characteristic analysis process of the vibration of the vibration sensor indicated by the detection result, and performs an abnormality determination process of the production facility using a threshold value,
Further comprising an extracting unit for extracting the data determined in the state analysis process that the production facility is not in a normal state,
The generation means receives correction information based on the extracted data, updates the threshold value,
The said correction information is an analysis apparatus containing the information which linked|related the malfunction event and the vibration characteristic. 　生産設備に設けられた振動センサの検出結果に基づいて当該生産設備の異常判定処理を第１の判別器を用いて行う判定手段と、
　前記第１の判別器により正常か異常かの判別ができなかった前記検出結果を画像化する画像処理手段と、
　前記画像化したデータを機械学習処理の対象として第２の判別器を生成する生成手段と、
　前記第２の判別器を用いて前記生産設備の状態解析処理を行う解析手段と、
を備える、
解析装置。 Determination means for performing abnormality determination processing of the production facility using the first discriminator based on the detection result of the vibration sensor provided in the production facility;
Image processing means for imaging the detection result, which cannot be discriminated as normal or abnormal by the first discriminator;
Generating means for generating a second discriminator using the imaged data as a target of machine learning processing;
Analysis means for performing a state analysis process of the production facility using the second discriminator;
With
Analyzer. 　請求項３に記載の解析装置において、
　前記状態解析処理で前記生産設備が正常状態でないと判定されたデータを抽出する抽出手段をさらに備え、
　前記生成手段は、抽出された前記データに基づく補正情報を受け付け前記第１の判別器を更新し、
　前記補正情報は、不具合事象と振動特性とを紐付けた情報を含む、解析装置。 The analysis device according to claim 3,
The state analysis process further comprises an extracting unit for extracting data determined to be in a non-normal state of the production facility,
The generation means receives correction information based on the extracted data, updates the first discriminator,
The said correction information is an analysis apparatus containing the information which linked|related the malfunction event and the vibration characteristic. 　請求項１から４のいずれか一項に記載の解析装置において、
　前記生成手段は、前記状態解析処理により正常と判定された前記データを前記機械学習処理の教師データとする、解析装置。 The analysis device according to any one of claims 1 to 4,
The analysis device, wherein the generation unit uses the data determined to be normal by the state analysis process as teacher data for the machine learning process. 　請求項１から５のいずれか一項に記載の解析装置において、
　前記検出結果に対してノイズ除去処理を行う処理手段をさらに備え、
　前記画像処理手段は、前記処理手段による前記ノイズ除去処理が行われた後の前記検出結果を画像化する、解析装置。 The analysis device according to any one of claims 1 to 5,
Further comprising processing means for performing noise removal processing on the detection result,
The analysis device, wherein the image processing means images the detection result after the noise removal processing is performed by the processing means. 　請求項１から６のいずれか一項に記載の解析装置において、
　前記生産設備は、ベルトコンベアであり、
　前記振動センサは、前記ベルトコンベアに設けられた複数の振動センサである、解析装置。 The analysis device according to any one of claims 1 to 6,
The production facility is a belt conveyor,
The said vibration sensor is an analysis apparatus which is a some vibration sensor provided in the said belt conveyor. 　解析装置が、
　生産設備に設けられた振動センサの検出結果を画像化し、
　前記画像化したデータを機械学習処理の対象にして判別器を生成し、
　前記判別器を用いて前記生産設備の状態解析処理を行う、
　解析方法。 The analysis device
Visualize the detection result of the vibration sensor installed in the production facility,
Generate a discriminator with the imaged data as the target of machine learning processing,
Performing a state analysis process of the production facility using the discriminator,
analysis method. 　請求項８に記載の解析方法において、
　前記解析装置が、さらに、
　前記検出結果が示す前記振動センサの振動の特徴解析処理を行い、閾値を用いて前記生産設備の異常判定処理を行い、
　前記状態解析処理で前記生産設備が正常状態でないと判定されたデータを抽出し、
　抽出された前記データに基づく補正情報を受け付け、前記閾値を更新し、
　前記補正情報は、不具合事象と振動特性とを紐付けた情報を含む、解析方法。 In the analysis method according to claim 8,
The analysis device further comprises
Performing a characteristic analysis process of the vibration of the vibration sensor indicated by the detection result, performing an abnormality determination process of the production facility using a threshold,
Extracting the data determined that the production equipment is not in a normal state in the state analysis process,
Accept correction information based on the extracted data, update the threshold,
The said correction information is an analysis method containing the information which linked|related the malfunction event and the vibration characteristic. 　解析装置が、
　生産設備に設けられた振動センサの検出結果に基づいて当該生産設備の異常判定処理を第１の判別器を用いて行い、
　前記第１の判別器により正常か異常かの判別ができなかった前記検出結果を画像化し、
　前記画像化したデータを機械学習処理の対象として第２の判別器を生成し、
　前記第２の判別器を用いて前記生産設備の状態解析処理を行う、
　解析方法。 The analysis device
Based on the detection result of the vibration sensor provided in the production equipment, the abnormality determination processing of the production equipment is performed using the first discriminator,
The detection result, which could not be discriminated as normal or abnormal by the first discriminator, is imaged,
A second discriminator is generated by using the imaged data as a target of machine learning processing,
Performing a state analysis process of the production facility using the second discriminator,
analysis method. 　請求項１０に記載の解析方法において、
　前記解析装置が、さらに、
　前記状態解析処理で前記生産設備が正常状態でないと判定されたデータを抽出し、
　抽出された前記データに基づく補正情報を受け付け前記第１の判別器を更新し、
　前記補正情報は、不具合事象と振動特性とを紐付けた情報を含む、解析方法。 The analysis method according to claim 10,
The analysis device further comprises
Extracting the data determined that the production equipment is not in a normal state in the state analysis process,
Accepting correction information based on the extracted data, updating the first discriminator,
The said correction information is an analysis method containing the information which linked|related the malfunction event and the vibration characteristic. 　請求項８から１１のいずれか一項に記載の解析方法において、
　前記解析装置が、さらに、
　前記状態解析処理により正常と判定された前記データを前記機械学習処理の教師データとする、解析方法。 The analysis method according to any one of claims 8 to 11,
The analysis device further comprises
An analysis method, wherein the data determined to be normal by the state analysis process is used as teacher data for the machine learning process. 　請求項８から１２のいずれか一項に記載の解析方法において、
　前記解析装置が、さらに、
　前記検出結果に対してノイズ除去処理を行い、
　前記ノイズ除去処理が行われた後の前記検出結果を画像化する、解析方法。 The analysis method according to any one of claims 8 to 12,
The analysis device further comprises
Noise removal processing is performed on the detection result,
An analysis method of imaging the detection result after the noise removal processing is performed. 　請求項８から１３のいずれか一項に記載の解析方法において、
　前記生産設備は、ベルトコンベアであり、
　前記振動センサは、前記ベルトコンベアに設けられた複数の振動センサである、解析方法。 The analysis method according to any one of claims 8 to 13,
The production facility is a belt conveyor,
An analysis method, wherein the vibration sensor is a plurality of vibration sensors provided on the belt conveyor. 　コンピュータに、
　生産設備に設けられた振動センサの検出結果を画像化する手順、
　前記画像化したデータを機械学習処理の対象にして判別器を生成する手順、
　前記判別器を用いて前記生産設備の異常判定処理を行う手順、を実行させるためのプログラム。 On the computer,
Procedure for imaging the detection result of the vibration sensor installed in the production facility,
A procedure for generating a discriminator by subjecting the imaged data to machine learning processing,
A program for executing a procedure of performing abnormality determination processing of the production facility using the discriminator. 　請求項１５に記載のプログラムにおいて、
　前記検出結果が示す前記振動センサの振動の特徴解析処理を行い、閾値を用いて前記生産設備の異常判定処理を行う手順、
　前記状態解析処理で前記生産設備が正常状態でないと判定されたデータを抽出する手順、
　前記生成する手順において、抽出された前記データに基づく補正情報を受け付け、前記閾値を更新する手順、をさらにコンピュータに実行させ、
　前記補正情報は、不具合事象と振動特性とを紐付けた情報を含む、プログラム。 The program according to claim 15,
Performing a characteristic analysis process of the vibration of the vibration sensor indicated by the detection result, a procedure for performing an abnormality determination process of the production facility using a threshold,
A procedure for extracting data that is determined in the state analysis process that the production facility is not in a normal state,
In the step of generating, the correction information based on the extracted data is accepted, and the procedure of updating the threshold value is further executed by a computer,
The said correction information is a program containing the information which linked|related the malfunction event and the vibration characteristic. 　コンピュータに、
　生産設備に設けられた振動センサの検出結果に基づいて当該生産設備の異常判定処理を第１の判別器を用いて行う手順、
　前記第１の判別器により正常か異常かの判別ができなかった前記検出結果を画像化する手順、
　前記画像化したデータを機械学習処理の対象として第２の判別器を生成する手順、
　前記第２の判別器を用いて前記生産設備の状態解析処理を行う手順、を実行させるためのプログラム。 On the computer,
A procedure for performing abnormality determination processing of the production facility using the first discriminator based on the detection result of the vibration sensor provided in the production facility,
A procedure for imaging the detection result, which cannot be discriminated as normal or abnormal by the first discriminator,
A procedure for generating a second discriminator using the imaged data as a target of machine learning processing;
A program for executing a procedure for performing a state analysis process of the production facility using the second discriminator. 　請求項１７に記載のプログラムにおいて、
　前記状態解析処理で前記生産設備が正常状態でないと判定されたデータを抽出する手順、
　前記生成する手順において、抽出された前記データに基づく補正情報を受け付け前記第１の判別器を更新する手順、をさらにコンピュータに実行させ、
　前記補正情報は、不具合事象と振動特性とを紐付けた情報を含む、プログラム。 The program according to claim 17,
A procedure for extracting data that is determined in the state analysis process that the production facility is not in a normal state,
In the step of generating, further causing a computer to execute a step of receiving correction information based on the extracted data and updating the first discriminator,
The said correction information is a program containing the information which linked|related the malfunction event and the vibration characteristic. 　請求項１５から１８のいずれか一項に記載のプログラムにおいて、
　前記生成する手順において、前記状態解析処理により正常と判定された前記データを前記機械学習処理の教師データとする手順をさらにコンピュータに実行させるためのプログラム。 The program according to any one of claims 15 to 18,
A program for causing a computer to further execute a procedure of using the data determined to be normal by the state analysis processing as teacher data of the machine learning processing in the generation procedure. 　請求項１５から１９のいずれか一項に記載のプログラムにおいて、
　前記検出結果に対してノイズ除去処理を行う手順、
　前記画像化する手順において、前記ノイズ除去処理が行われた後の前記検出結果を画像化する手順、をさらにコンピュータに実行させるためのプログラム。 The program according to any one of claims 15 to 19,
A procedure for performing noise removal processing on the detection result,
A program for causing a computer to further execute the step of imaging the detection result after the noise removal processing is performed in the step of imaging. 　請求項１５から２０のいずれか一項に記載のプログラムにおいて、
　前記生産設備は、ベルトコンベアであり、
　前記振動センサは、前記ベルトコンベアに設けられた複数の振動センサである、プログラム。 The program according to any one of claims 15 to 20,
The production facility is a belt conveyor,
A program in which the vibration sensor is a plurality of vibration sensors provided on the belt conveyor.
PCT/JP2020/004042 2019-02-05 2020-02-04 Analysis device, analysis method, and program Ceased WO2020162425A1 (en)

Priority Applications (4)

Application Number Priority Date Filing Date Title
KR1020217024139A KR20210107844A (en) 2019-02-05 2020-02-04 Analysis apparatus, analysis method, and program
PH1/2021/551623A PH12021551623A1 (en) 2019-02-05 2020-02-04 Analysis apparatus, analysis method, and program
CN202080012631.9A CN113383216A (en) 2019-02-05 2020-02-04 Analysis device, analysis method, and program
JP2020571198A JP7188463B2 (en) 2019-02-05 2020-02-04 ANALYSIS DEVICE, ANALYSIS METHOD, AND PROGRAM

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2019019068 2019-02-05
JP2019-019068 2019-02-05

Publications (1)

Publication Number Publication Date
WO2020162425A1 true WO2020162425A1 (en) 2020-08-13

Family

ID=71947695

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2020/004042 Ceased WO2020162425A1 (en) 2019-02-05 2020-02-04 Analysis device, analysis method, and program

Country Status (6)

Country Link
JP (1) JP7188463B2 (en)
KR (1) KR20210107844A (en)
CN (1) CN113383216A (en)
PH (1) PH12021551623A1 (en)
TW (1) TW202045898A (en)
WO (1) WO2020162425A1 (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2022064769A1 (en) * 2020-09-24 2022-03-31 国立大学法人大阪大学 Degradation-status prediction system and degradation-status prediction method
WO2022158073A1 (en) * 2021-01-25 2022-07-28 株式会社日本製鋼所 Computer program, abnormality detection method, abnormality detection device, molding machine system, and learning model generation method
JP2023053639A (en) * 2021-10-01 2023-04-13 株式会社デンソー Noise factor determination device and noise factor determination method
JP2023083737A (en) * 2021-12-06 2023-06-16 Ｉｈｉ運搬機械株式会社 Belt conveyor abnormality detection device

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2008232708A (en) * 2007-03-19 2008-10-02 Jfe Steel Kk Deterioration degree diagnosis method, deterioration degree diagnosis device, and deterioration diagnosis program
JP2018178810A (en) * 2017-04-10 2018-11-15 株式会社デンソーテン Knock control device, knock adaptation method and knock adaptation program
WO2018216258A1 (en) * 2017-05-25 2018-11-29 日本電気株式会社 Processing device, processing method, and program

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP3170076B2 (en) * 1992-12-18 2001-05-28 株式会社小野測器 Rolling bearing failure diagnosis device
JP2000321176A (en) * 1999-05-17 2000-11-24 Mitsui Eng & Shipbuild Co Ltd Error detection method and device
JP2011107093A (en) * 2009-11-20 2011-06-02 Jx Nippon Oil & Energy Corp Apparatus and method for diagnosing abnormality of vibrating body
CN102494882B (en) * 2011-11-30 2013-11-06 中国神华能源股份有限公司 Device for online monitoring and fault diagnosis of spring of vibrating screen for mining and method thereof
JP6877978B2 (en) * 2016-12-06 2021-05-26 日本電気通信システム株式会社 Learning equipment, learning methods and programs
CN107560849B (en) * 2017-08-04 2020-02-18 华北电力大学 Wind turbine bearing fault diagnosis method based on multi-channel deep convolutional neural network
CN108896296A (en) * 2018-04-18 2018-11-27 北京信息科技大学 A kind of wind turbine gearbox method for diagnosing faults based on convolutional neural networks

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2008232708A (en) * 2007-03-19 2008-10-02 Jfe Steel Kk Deterioration degree diagnosis method, deterioration degree diagnosis device, and deterioration diagnosis program
JP2018178810A (en) * 2017-04-10 2018-11-15 株式会社デンソーテン Knock control device, knock adaptation method and knock adaptation program
WO2018216258A1 (en) * 2017-05-25 2018-11-29 日本電気株式会社 Processing device, processing method, and program

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2022064769A1 (en) * 2020-09-24 2022-03-31 国立大学法人大阪大学 Degradation-status prediction system and degradation-status prediction method
WO2022158073A1 (en) * 2021-01-25 2022-07-28 株式会社日本製鋼所 Computer program, abnormality detection method, abnormality detection device, molding machine system, and learning model generation method
JP2022113523A (en) * 2021-01-25 2022-08-04 株式会社日本製鋼所 Computer program, anomaly detection method, anomaly detection device, molding machine system, and learning model generation method
JP7581063B2 (en) 2021-01-25 2024-11-12 株式会社日本製鋼所 COMPUTER PROGRAM, ABNORMALITY DETECTION METHOD, ABNORMALITY DETECTION DEVICE, MOLDING MACHINE SYSTEM, AND LEARNING MODEL GENERATION METHOD
JP2025003752A (en) * 2021-01-25 2025-01-09 株式会社日本製鋼所 COMPUTER PROGRAM, ABNORMALITY DETECTION METHOD, ABNORMALITY DETECTION DEVICE, MOLDING MACHINE SYSTEM, AND LEARNING MODEL GENERATION METHOD
JP2023053639A (en) * 2021-10-01 2023-04-13 株式会社デンソー Noise factor determination device and noise factor determination method
JP7647480B2 (en) 2021-10-01 2025-03-18 株式会社デンソー Noise factor determination device and noise factor determination method
JP2023083737A (en) * 2021-12-06 2023-06-16 Ｉｈｉ運搬機械株式会社 Belt conveyor abnormality detection device

Also Published As

Publication number Publication date
JP7188463B2 (en) 2022-12-13
JPWO2020162425A1 (en) 2021-12-09
PH12021551623A1 (en) 2022-04-18
KR20210107844A (en) 2021-09-01
CN113383216A (en) 2021-09-10
TW202045898A (en) 2020-12-16

Similar Documents

Publication Publication Date Title
RU2704073C2 (en) Method and system for training acoustic or vibration analysis of machine
WO2020162425A1 (en) Analysis device, analysis method, and program
CN107272586B (en) Machine learning device, machine learning method, failure prediction device, and failure prediction system
JP6837848B2 (en) Diagnostic device
US20180264613A1 (en) Abnormality detection apparatus and machine learning apparatus
EP3206103A1 (en) Model based system monitoring
JP6200833B2 (en) Diagnostic equipment for plant and control equipment
JP7006282B2 (en) Equipment abnormality diagnostic equipment
US11604458B2 (en) Anomaly determination device and anomaly determination method
KR20210006832A (en) Method and apparatus for machine fault diagnosis
WO2014031291A2 (en) Fleet anomaly detection system and method
TWI861824B (en) Abnormal diagnosis method, abnormal diagnosis device, and abnormal diagnosis program
JP6898607B2 (en) Abnormality sign detection system and abnormality sign detection method
CN113591984A (en) Method and device for detecting equipment operation event, electronic equipment and storage medium
JP2002268742A (en) Working diagnosis monitoring system and its device and its program
JP2022082208A (en) Pattern classification device, elevator sound diagnosis system, pattern classification method, elevator sound diagnosis device, and elevator sound diagnosis method
JP2024109377A (en) Vehicle Diagnostic System
Jalayer et al. Evaluating deep learning models for fault diagnosis of a rotating machinery with epistemic and aleatoric uncertainty
JP2004020484A (en) Error monitoring device and error monitoring program
CN121219561A (en) Model of soft sensor for measuring mechanical damage and soft sensor
CN110411730A (en) A kind of rotating equipment failures judgment method, system and readable storage medium storing program for executing
Tastimur et al. Defect diagnosis of rolling element bearing using deep learning
KR20230127535A (en) Apparatus for riveting defect diagnosis based on machine learning and method thereof
Hickenbottom Proactive approaches for Engine Health Management and a high value example
Martiri et al. A Novel Benchmark for Fault Detection in Rolling Bearings Using CNNs and Monte Carlo Dropout

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20751935

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 20217024139

Country of ref document: KR

Kind code of ref document: A

ENP Entry into the national phase

Ref document number: 2020571198

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20751935

Country of ref document: EP

Kind code of ref document: A1