JP7776973B2 - Condition monitoring device and condition monitoring method - Google Patents

Description

This disclosure relates to a condition monitoring device and a condition monitoring method.

In recent years, railway vehicle maintenance systems have been shifting from time-based maintenance to condition-based maintenance. Changing maintenance systems requires the introduction of condition monitoring technology, and various research and development efforts are underway. A conventional condition monitoring device is the condition diagnosis device described in Patent Document 1. This conventional condition diagnosis device is configured to diagnose the condition of traction equipment installed on railway vehicles. The device comprises: an aggregation means that aggregates the results of octave-band analysis of vibration data detected by a vibration sensor for detecting vibrations in the traction equipment installed on the railway vehicle, by operating mode and running speed of the railway vehicle, or by running speed; and a discrimination means that determines whether the aggregation results of the aggregation means match the aggregation result reference data for each of N (N≧1) types of known states of the traction equipment.

Japanese Patent Application Laid-Open No. 2017-77055

In conventional methods of monitoring the condition of an object based on the results of vibration detection, it is necessary to use complex algorithms to capture the vibration frequency characteristics and vibration changes for each item being monitored. When there are many items being monitored, installing vibration sensors and adding algorithms for each item can lead to complex configurations and processing. When configurations and processing become more complex, the operation and management of the device itself becomes more cumbersome, and reducing the labor required for maintenance becomes a new challenge.

This disclosure has been made to solve the above-mentioned problems, and aims to provide a condition monitoring device and a condition monitoring method that can monitor the status of a large number of target items without complicating the configuration and processing.

A condition monitoring device according to one aspect of the present disclosure includes a vibration sensor installed on a railway vehicle, a determination unit that determines whether or not an abnormality exists for each condition monitoring target item based on a detection signal from the vibration sensor, and a memory unit that stores a lookup table that associates a calculation interval for the detection signal with each target item. The determination unit calculates the interval sum of squares of the output value of the detection signal based on the calculation interval for each target item, and determines whether or not an abnormality exists for each target item by comparing the interval sum of squares with a preset threshold.

This condition monitoring device calculates the interval sum of squares for the output value of the detection signal from the vibration sensor installed on the railway vehicle, and determines whether or not an abnormality exists for each target item by comparing the interval sum of squares with a threshold. With this method, by adjusting the calculation interval for calculating the sum of squares, it is possible to extract multiple feature quantities corresponding to each target item from the same detection signal. Therefore, this condition monitoring device does not need to install a vibration sensor for each target item or add complex algorithms, and can monitor the status of multiple target items without complicating the configuration or processing.

The reference table may include target items associated with a first calculation interval that includes the entire detection signal interval, and target items associated with a second calculation interval that is shorter than the first calculation interval. Target items for condition monitoring in railway systems may include, for example, vehicle running systems such as wheels and dampers, and infrastructure systems such as rails and bridges. These target items may be divided into items that should be monitored for long-term abnormalities and items that should be monitored for short-term abnormalities. Therefore, by associating the calculation interval for calculating the sum of squares with both short-term and long-term abnormality monitoring, condition monitoring that is more in line with the actual state of the railway system can be performed.

In the lookup table, target items may also be associated with vibration direction, frequency band, and sampling frequency. This allows multiple feature amounts corresponding to each target item to be extracted more accurately from the same detection signal.

The vibration sensor may be a three-axis acceleration sensor. This further reduces the number of vibration sensors that need to be installed.

Vibration sensors may be installed at each of the front and rear ends of the railway vehicle, or on each of the front and rear bogies. By installing vibration sensors at these locations on the railway vehicle, vibrations related to both the running system and the track system of the railway system can be efficiently acquired. Furthermore, by extracting feature values from the detection signals from vibration sensors at different locations on the front and rear of the railway vehicle, the accuracy of determining whether or not an abnormality exists can be improved.

A condition monitoring method according to one aspect of the present disclosure includes a determination step of determining whether or not an abnormality exists for each condition monitoring target item based on a detection signal from a vibration sensor. In the determination step, a lookup table that associates a calculation interval for the detection signal with each target item is referenced, the interval sum of squares of the output value of the detection signal is calculated based on the calculation interval for each target item, and the interval sum of squares is compared with a preset threshold value to determine whether or not an abnormality exists for each target item.

This condition monitoring method calculates the interval sum of squares for the output value of the detection signal from a vibration sensor installed on the railway vehicle, and determines whether or not an abnormality exists for each target item by comparing the interval sum of squares with a threshold. With this method, by adjusting the calculation interval for calculating the sum of squares, it is possible to extract multiple feature quantities corresponding to each target item from the same detection signal. Therefore, this condition monitoring method does not require the installation of a vibration sensor or the addition of an algorithm for each target item, and can monitor the status of multiple target items without complicating the configuration or processing.

This disclosure makes it possible to monitor the status of a large number of target items without complicating the configuration and processing.

1 is a schematic configuration diagram illustrating an embodiment of a condition monitoring device according to the present disclosure. FIG. 10 is a diagram illustrating an example of a reference table relating to target items associated with a first calculation interval. FIG. 10 is a diagram illustrating an example of a reference table relating to target items associated with second calculation intervals. 2 is a flowchart showing an example of the operation of the state monitoring device shown in FIG. 1 . FIG. 2 is a perspective view showing the configuration of a simulated vehicle and a simulated rail used in the examples. 10 is a diagram showing an example of a detection signal from a vibration sensor when a derailment occurs in a wheel on the rear side in the traveling direction of the simulated vehicle. FIG. FIG. 7 is a diagram showing the section sum of squares of the detection signal shown in FIG. 6 . FIG. 10 is a diagram showing an example of a detection signal from a vibration sensor when a tread flat occurs on a wheel on the rear side in the traveling direction of a simulated vehicle. FIG. 9 is a diagram showing the section sum of squares of the detection signal shown in FIG. 8 . 10 is a diagram showing an example of a detection signal from a vibration sensor when a wheel on the rear side in the traveling direction of the simulated vehicle is causing a snake motion. FIG. FIG. 11 is a diagram showing the section sum of squares of the detection signal shown in FIG. 10 .

Below, preferred embodiments of a condition monitoring device and a condition monitoring method according to one aspect of the present disclosure will be described in detail with reference to the drawings.

Figure 1 is a schematic diagram illustrating one embodiment of a condition monitoring device according to the present disclosure. This condition monitoring device 1 is configured as a device that monitors the condition of a railway vehicle system, and performs condition monitoring of vehicle running systems such as wheels and dampers, and infrastructure systems such as rails and bridges. As shown in Figure 1, the condition monitoring device 1 includes a vibration sensor 2, a determination unit 3, and a memory unit 4.

The vibration sensor 2 is, for example, a three-axis acceleration sensor that detects acceleration in the X, Y, and Z directions of an object. The vibration sensor 2 is provided on the railway vehicle 11. The vibration sensor 2 may also be provided on a bogie of the railway vehicle 11. In this embodiment, the vibration sensor 2 is provided on each of the front and rear bogies 12, 12 of the railway vehicle 11. In this case, each of the vibration sensors 2 may be located in an area directly above the center plate of the bogie when viewed from the height direction of the railway vehicle 11. The vibration sensor 2 detects vibrations (acceleration) applied to the bogie 12 of the railway vehicle 11 and sequentially outputs detection signals to the determination unit 3.

The judgment unit 3 is a part that judges whether or not there is an abnormality for each item subject to status monitoring based on the detection signal from the vibration sensor 2. Physically, the judgment unit 3 is a computer system equipped with memory such as RAM and ROM, a processor (arithmetic circuit) such as a CPU, a communications interface, and a storage unit such as a hard disk. Examples of such computer systems include personal computers, cloud servers, and smart devices (smartphones, tablet terminals, etc.). The computer system functions as the judgment unit 3 by having the CPU execute a program stored in the memory. The judgment unit 3 is located, for example, in a maintenance center for the railway vehicle system, and is capable of receiving detection signals from the vibration sensor 2 via the network N.

When the judgment unit 3 receives a detection signal from the vibration sensor 2, it calculates the interval sum of squares of the output value of the detection signal based on the calculation interval for each target item of status monitoring. It then compares the interval sum of squares with a preset threshold to determine whether or not there is an abnormality for each target item. The judgment unit 3 determines whether or not there is an abnormality for each target item, for example, based on whether or not the interval sum of squares calculated using a specified calculation interval exceeds the threshold. If the judgment unit 3 determines that there is an abnormality in the target item, it transmits a signal indicating that there is an abnormality in the target item to, for example, the running control device of the railway vehicle 11 or a control device at the railway system maintenance center.

The interval sum of squares is a value obtained by adding up the squares of data for a certain interval, including that data. When calculating the interval sum of squares, the judgment unit 3 references a lookup table 5 stored in the memory unit 4. The lookup table 5 associates a calculation interval for the detection signal with each target item of status monitoring. In this embodiment, the vibration sensor 2 repeatedly outputs a detection signal, with one detection cycle lasting 180 seconds. As shown in FIG. 2, the lookup table 5 includes target items associated with a first calculation interval that includes the entire interval of the detection signal (the entire period of one detection cycle). As shown in FIG. 3, the lookup table 5 also includes target items associated with a second calculation interval that is shorter than the first calculation interval. The target items shown in FIG. 2 are items that should be monitored for abnormalities over the long term, while the target items shown in FIG. 3 are items that should be monitored for abnormalities over the short term.

In reference table 5, "vibration direction," "frequency band," "sampling frequency," and "threshold" are associated with each target item. "Vibration direction" is a parameter that indicates which direction of acceleration detection signal to use from the detection signals output from vibration sensor 2, which is a three-axis acceleration sensor. "Frequency band" is a parameter that indicates which frequency band of signal components to use from the detection signals output from vibration sensor 2. "Sampling frequency" is a parameter that indicates how many digital data pieces per second the detection signal is converted into. "Threshold" is a parameter that is set for the interval squared sum of the detection signal. The "threshold" may be a value normalized based on the maximum value of the detection signal.

In the example in Figure 2, three items have been set as targets for long-term monitoring: "air spring abnormality," "axle spring/bearing abnormality," and "track abnormality." For "air spring abnormality," the "vibration direction" is up and down, the "frequency band" is 5-20 Hz, the "calculation interval" is 180,000 data points (180 seconds per entire interval), the "sampling frequency" is 1 kHz, and the "threshold" is 3,000. For "axle spring/bearing abnormality," the "vibration direction" is up and down, the "frequency band" is 5-20 Hz, the "calculation interval" is 180,000 data points (180 seconds per entire interval), the "sampling frequency" is 1 kHz, and the "threshold" is 2,000. For "track abnormality," the "vibration direction" is up and down, the "frequency band" is 5-20 Hz, the "calculation interval" is 180,000 data points (180 seconds per entire interval), the "sampling frequency" is 1 kHz, and the "threshold" is 2,000.

In the example shown in Figure 2, the parameters for "axle spring/bearing abnormality" and "track abnormality" are the same, but in this case, although not shown, an additional parameter, the length of time during which the interval square sum calculated using a specified calculation interval exceeds a threshold, can be introduced to distinguish between the two.

In the example in Figure 3, seven items have been set as targets for short-term monitoring: "derailment," "flat wheel tread," "snake motion," "lateral damper abnormality/yaw damper abnormality," "axle damper abnormality," "rail corrugation," and "bridge abnormality." For "derailment," the "vibration direction" is up and down, the "frequency band" is overall (DC to 1 kHz), the "calculation interval" is 100 data (100 msec), the "sampling frequency" is 1 kHz, and the "threshold" is 2000.

For "flat wheel tread," the "vibration direction" is up and down, the "frequency band" is overall (DC to 1 kHz), the "calculation interval" is 1500 data (1.5 seconds), the "sampling frequency" is 1 kHz, and the "threshold" is 1500. For "snake motion," the "vibration direction" is left and right, the "frequency band" is overall (DC to 1 kHz), the "calculation interval" is 1500 data (1.5 seconds), the "sampling frequency" is 1 kHz, and the "threshold" is 1500.

For "Lateral Damper Abnormality/Yaw Damper Abnormality," the "Vibration Direction" is left and right, the "Frequency Band" is 5 to 20 Hz, the "Calculation Range" is 1000 data (1.0 sec), the "Sampling Frequency" is 1 kHz, and the "Threshold" is 1000. For "Axial Damper Abnormality," the "Vibration Direction" is up and down, the "Frequency Band" is 5 to 20 Hz, the "Calculation Range" is 1000 data (1.0 sec), the "Sampling Frequency" is 1 kHz, and the "Threshold" is 1000.

For "rail corrugated wear," the "vibration direction" is up and down, the "frequency band" is 5 to 20 Hz, the "calculation interval" is 1000 data (1.0 sec), the "sampling frequency" is 1 kHz, and the "threshold" is 1000. For "bridge abnormalities," the "vibration direction" is up and down, the "frequency band" is 13 to 15 Hz, the "calculation interval" is 1600 data (1.6 sec), the "sampling frequency" is 1 kHz, and the "threshold" is 1000.

In the example of Figure 3, the parameters for "axle damper abnormality" and "rail corrugation" are the same, but as with the "axle spring/bearing abnormality" and "track abnormality" in Figure 2, by introducing an additional parameter - the length of time during which the interval sum of squares calculated using a specified calculation interval exceeds a threshold - it is possible to distinguish between the two.

Figure 4 is a flowchart showing an example of the operation of the condition monitoring device 1 described above. The condition monitoring device 1 executes a judgment step to determine the presence or absence of an abnormality for each condition monitoring target item based on the detection signal from the vibration sensor 2. In the judgment step, as shown in Figure 4, first, the detection signal from the vibration sensor 2 is acquired (step S01). Next, the judgment unit 3 references a reference table 5 that associates calculation intervals for the detection signal with each target item (step S02). The judgment unit 3 calculates the interval sum of squares of the output value of the detection signal based on the calculation interval for each target item (step S03). After calculating the interval sum of squares, the calculated interval sum of squares is compared with a preset threshold value (step S04). The presence or absence of an abnormality for each target item is determined by comparing the interval sum of squares with the threshold value (step S05).

As explained above, the condition monitoring device 1 calculates the interval sum of squares for the output value of the detection signal from the vibration sensor 2 installed on the railway vehicle 11, and determines the presence or absence of an abnormality for each target item by comparing the interval sum of squares with a threshold value. With this method, by adjusting the calculation interval for calculating the sum of squares, it is possible to extract multiple feature quantities corresponding to each target item from the same detection signal. Therefore, the condition monitoring device 1 does not need to install a vibration sensor 2 for each target item or add complex algorithms, and can monitor the status of a large number of target items without complicating the configuration or processing.

In this embodiment, reference table 5 includes target items associated with a first calculation interval that includes the entire detection signal interval, and target items associated with a second calculation interval that is shorter than the first calculation interval. Target items for condition monitoring in railway systems may include, for example, vehicle running systems such as wheels and dampers, and infrastructure systems such as rails and bridges. These target items can be divided into items that should be monitored for long-term abnormalities and items that should be monitored for short-term abnormalities. Therefore, by associating the calculation interval for calculating the sum of squares with both short-term and long-term abnormality monitoring, condition monitoring that is more in line with the actual state of the railway system can be performed.

In this embodiment, in the lookup table 5, the target items may further be associated with vibration direction, frequency band, and sampling frequency. This allows multiple feature amounts corresponding to each target item to be extracted more accurately from the same detection signal.

In this embodiment, a three-axis acceleration sensor is used as the vibration sensor 2. This makes it possible to reduce the number of vibration sensors 2 that need to be installed. Furthermore, in this embodiment, a vibration sensor 2 is provided on each of the front and rear bogies 12, 12 of the railway vehicle. By providing the vibration sensor 2 on the bogie 12 of the railway vehicle 11, it is possible to preferably acquire vibrations related to both the running system and the installation system of the railway system. Furthermore, by extracting feature amounts from the detection signals from the vibration sensors 2 at different positions, it is possible to improve the accuracy of determining whether or not an abnormality exists.

The present disclosure is not limited to the above-described embodiment. For example, in the above-described embodiment, the determination unit 3 receives detection signals from the vibration sensors 2 via the network N. However, the determination unit 3 may be provided on the railway vehicle 11 (e.g., in the driver's cab) and receive detection signals from the vibration sensors 2 via a wired or wireless connection. Furthermore, the vibration sensors 2 do not necessarily have to be provided on each of the front and rear bogies 12, 12 of the railway vehicle 11, and may be provided on only one of the bogies 12. The location where the vibration sensors 2 are provided is not necessarily limited to the bogie 12, and they may be provided at each of the front and rear ends of the railway vehicle 11, for example. When the vibration sensors 2 are provided at the front and rear ends of the railway vehicle 11, the vibration sensors 2 may be located under the floor in an area that does not overlap with the bogie 12 when viewed from the height direction of the railway vehicle 11.

An example of condition monitoring is described below. In this example, a simulated vehicle 21 and simulated rail (not shown) as shown in Figure 5 were used, and detection signals were obtained by vibration sensors 2 and the section sum of squares of the detection signals was calculated for three of the condition monitoring target items shown in Figures 2 and 3: "derailment," "wheel tread flat," and "snake behavior." In the example of Figure 5, the vibration sensors 2 were placed near the center of the width of the simulated vehicle 21 (the center dish portion) on the front and rear bogie-equivalent portions 23, 23 of the simulated vehicle 21.

FIG. 6 shows an example of the detection signal from the vibration sensor when a derailment occurs in the rear wheel of the simulated vehicle. In FIG. 6, the horizontal axis represents time (sec) and the vertical axis represents acceleration (m/sec ² ), and the detection signal in the Z direction (vehicle height direction) from the vibration sensor 2 is plotted. Graph A in FIG. 6 is the detection signal from the vibration sensor 2 located on the bogie-equivalent portion 23 on the rear side of the simulated vehicle 21 in the traveling direction, and graph B in FIG. 6 is the detection signal from the vibration sensor 2 located on the bogie-equivalent portion 23 on the front side of the simulated vehicle 21 in the traveling direction. The sampling frequency is 1 kHz. In FIG. 6, it can be seen that the acceleration increases and decreases sharply from 2.8 seconds to 5.6 seconds after the start of measurement. Several small increases and decreases in acceleration can also be seen outside of this period.

Figure 7 is a diagram showing the interval sum of squares of the detection signal shown in Figure 6. In Figure 7, the horizontal axis shows time (sec) and the vertical axis shows the interval sum of squares (normalized value). Graphs A and B in Figure 7 are plots of the interval sum of squares of the detection signal corresponding to graphs A and B in Figure 6, respectively. The interval for calculating the interval sum of squares is 100 data points (1 msec). As shown in Figure 7, when the interval sum of squares of the detection signal is calculated, the waveform of the vibration component related to the derailment is extracted from the detection signal as a feature.

This vibration component waveform shows that a derailment actually occurred on the simulated vehicle 21 between 4.2 and 5.6 seconds after measurement began. Furthermore, comparing graphs A and B in Figure 7 reveals that, during the period 4.2 to 5.6 seconds after measurement began, the peak value and integral value of the waveform of the sum of squares of the section of the detection signal from the vibration sensor 2 located on the derailed bogie were greater than the peak value and integral value of the waveform of the sum of squares of the section of the detection signal from the vibration sensor 2 located on the non-derailed bogie. Therefore, by setting threshold values for at least one of the peak value and integral value for each of the two waveforms, it is possible to determine whether a derailment has occurred and which bogie's wheel has derailed.

FIG. 8 shows an example of the detection signal from the vibration sensor when a tread flat occurs on the wheel on the rear side of the simulated vehicle in the traveling direction. In FIG. 8, the horizontal axis represents time (sec) and the vertical axis represents acceleration (m/sec ² ), and the detection signal in the Z direction (vehicle height direction) from the vibration sensor 2 is plotted. Graph A in FIG. 8 is the detection signal from the vibration sensor 2 disposed on the bogie-equivalent portion 23 on the rear side of the simulated vehicle 21 in the traveling direction, and graph B in FIG. 8 is the detection signal from the vibration sensor 2 disposed on the bogie-equivalent portion 23 on the front side of the simulated vehicle 21 in the traveling direction. The sampling frequency is 1 kHz. In FIG. 8, it can be seen that the acceleration increases and decreases sharply from 2 seconds to 14 seconds after the start of measurement.

Figure 9 is a diagram showing the interval sum of squares of the detection signal shown in Figure 8. In Figure 9, the horizontal axis shows time (sec) and the vertical axis shows the interval sum of squares (normalized value). Graphs A and B in Figure 9 are plots of the interval sum of squares of the detection signal corresponding to graphs A and B in Figure 8, respectively. The calculation interval for the interval sum of squares is 1,500 data points (1.5 sec). As shown in Figure 9, when the interval sum of squares of the detection signal is calculated, the waveform of the vibration component related to the tread flat is extracted from the detection signal as a feature.

The waveform of this vibration component shows that a tread flat actually occurred on the simulated vehicle 21 between 2 and 12 seconds after measurement began. Furthermore, comparing graphs A and B in Figure 9 reveals that there are differences in the shapes of the two waveforms. Therefore, by setting threshold values such as peak and integral values for the two waveforms, it is possible to determine whether a tread flat occurred and on which bogie wheel the tread flat occurred.

Figure 10 is a diagram showing an example of the detection signal from the vibration sensor when the rear wheels of the simulated vehicle start to snake. In Figure 10, the horizontal axis represents time (sec) and the vertical axis represents acceleration (m/sec ² ), and the detection signal in the Y direction (vehicle width direction) from the vibration sensor 2 disposed on the bogie-equivalent portion 23 on the rear side of the simulated vehicle 21 is plotted. The sampling frequency is 1 kHz. Figure 10 shows that the acceleration increases and decreases sharply over a period of 10 seconds from the start of measurement.

Figure 11 shows the interval sum of squares of the detection signal shown in Figure 10. In Figure 11, the horizontal axis represents time (sec) and the vertical axis represents the interval sum of squares (normalized value). The calculation interval for the interval sum of squares is 1,500 data points (1.5 sec). As shown in Figure 11, when the interval sum of squares of the detection signal is calculated, the waveform of the vibration component related to hunting is extracted from the detection signal as a feature. In Figure 10, the acceleration appears to increase and decrease uniformly from immediately after the start of measurement through the 10-second period, but in Figure 11, the increase and decrease in the interval sum of squares is clearly evident, and it can be seen that the waveform peaks around 5.5 seconds after the start of measurement. Therefore, by setting a threshold for the peak value of this interval sum of squares waveform, it is possible to determine whether hunting is occurring.

In the detection signals during abnormal conditions shown in Figures 6, 8, and 10, a sudden increase or decrease in acceleration continues in response to the occurrence of the abnormality. However, the waveforms of the detection signals during the period when the abnormality occurs are similar to each other, making it difficult to identify the feature quantities of each target item for condition monitoring from these waveforms. On the other hand, the section sums of squares of the detection signals shown in Figures 7, 9, and 11 show that multiple feature quantities corresponding to each target item can be extracted as waveforms. Therefore, by setting appropriate thresholds for these section sums of squares, it is possible to monitor the condition of multiple target items without installing vibration sensors for each target item or adding complex algorithms.

1...Condition monitoring device, 2...Vibration sensor, 3...Determination unit, 4...Memory unit, 5...Reference table, 11...Railway vehicle, 12...Bogie.

Claims (6)

a vibration sensor provided on the railway vehicle;
a determination unit that determines whether or not there is an abnormality for each item subject to status monitoring based on the detection signal from the vibration sensor;
a storage unit that stores a reference table that associates a calculation interval for the detection signal with each of the target items,
The judgment unit calculates an interval sum of squares of the output values of the detection signal without performing filter processing based on the calculation interval for each of the target items, and determines whether or not there is an abnormality for each of the target items by comparing the interval sum of squares with a preset threshold value. A condition monitoring device as described in claim 1, wherein the reference table includes target items associated with a first calculation interval that includes the entire interval of the detection signal, and target items associated with a second calculation interval that is shorter than the first calculation interval. A condition monitoring device according to claim 1 or 2, wherein in the reference table, the target items are further associated with vibration direction, frequency band, and sampling frequency. A condition monitoring device according to any one of claims 1 to 3, wherein the vibration sensor is a three-axis acceleration sensor. A condition monitoring device according to any one of claims 1 to 4, wherein the vibration sensors are provided at each of the front and rear ends of the railway vehicle or each of the front and rear bogies. a determining step of determining whether or not there is an abnormality for each item to be monitored based on a detection signal from the vibration sensor;
In the judgment step, a reference table that associates a calculation interval for the detection signal with each target item is referenced, an interval sum of squares of the output value of the detection signal without filter processing is calculated based on the calculation interval for each target item, and the interval sum of squares is compared with a preset threshold value to determine whether or not there is an abnormality for each target item.