WO2019181728A1 - 劣化推定装置、コンピュータプログラム及び劣化推定方法 - Google Patents
劣化推定装置、コンピュータプログラム及び劣化推定方法 Download PDFInfo
- Publication number
- WO2019181728A1 WO2019181728A1 PCT/JP2019/010543 JP2019010543W WO2019181728A1 WO 2019181728 A1 WO2019181728 A1 WO 2019181728A1 JP 2019010543 W JP2019010543 W JP 2019010543W WO 2019181728 A1 WO2019181728 A1 WO 2019181728A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- time point
- soh
- soc
- storage element
- representative value
- 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
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/367—Software therefor, e.g. for battery testing using modelling or look-up tables
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/374—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC] with means for correcting the measurement for temperature or ageing
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/382—Arrangements for monitoring battery or accumulator variables, e.g. SoC
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/392—Determining battery ageing or deterioration, e.g. state of health
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01M—PROCESSES OR MEANS, e.g. BATTERIES, FOR THE DIRECT CONVERSION OF CHEMICAL ENERGY INTO ELECTRICAL ENERGY
- H01M10/00—Secondary cells; Manufacture thereof
- H01M10/05—Accumulators with non-aqueous electrolyte
- H01M10/052—Li-accumulators
- H01M10/0525—Rocking-chair batteries, i.e. batteries with lithium insertion or intercalation in both electrodes; Lithium-ion batteries
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01M—PROCESSES OR MEANS, e.g. BATTERIES, FOR THE DIRECT CONVERSION OF CHEMICAL ENERGY INTO ELECTRICAL ENERGY
- H01M10/00—Secondary cells; Manufacture thereof
- H01M10/42—Methods or arrangements for servicing or maintenance of secondary cells or secondary half-cells
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01M—PROCESSES OR MEANS, e.g. BATTERIES, FOR THE DIRECT CONVERSION OF CHEMICAL ENERGY INTO ELECTRICAL ENERGY
- H01M10/00—Secondary cells; Manufacture thereof
- H01M10/42—Methods or arrangements for servicing or maintenance of secondary cells or secondary half-cells
- H01M10/48—Accumulators combined with arrangements for measuring, testing or indicating the condition of cells, e.g. the level or density of the electrolyte
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01M—PROCESSES OR MEANS, e.g. BATTERIES, FOR THE DIRECT CONVERSION OF CHEMICAL ENERGY INTO ELECTRICAL ENERGY
- H01M10/00—Secondary cells; Manufacture thereof
- H01M10/42—Methods or arrangements for servicing or maintenance of secondary cells or secondary half-cells
- H01M10/48—Accumulators combined with arrangements for measuring, testing or indicating the condition of cells, e.g. the level or density of the electrolyte
- H01M10/486—Accumulators combined with arrangements for measuring, testing or indicating the condition of cells, e.g. the level or density of the electrolyte for measuring temperature
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J7/00—Circuit arrangements for charging or depolarising batteries or for supplying loads from batteries
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J7/00—Circuit arrangements for charging or depolarising batteries or for supplying loads from batteries
- H02J7/0047—Circuit arrangements for charging or depolarising batteries or for supplying loads from batteries with monitoring or indicating devices or circuits
- H02J7/0048—Detection of remaining charge capacity or state of charge [SOC]
- H02J7/0049—Detection of fully charged condition
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J7/00—Circuit arrangements for charging or depolarising batteries or for supplying loads from batteries
- H02J7/0047—Circuit arrangements for charging or depolarising batteries or for supplying loads from batteries with monitoring or indicating devices or circuits
- H02J7/005—Detection of state of health [SOH]
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01M—PROCESSES OR MEANS, e.g. BATTERIES, FOR THE DIRECT CONVERSION OF CHEMICAL ENERGY INTO ELECTRICAL ENERGY
- H01M2220/00—Batteries for particular applications
- H01M2220/20—Batteries in motive systems, e.g. vehicle, ship, plane
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E60/00—Enabling technologies; Technologies with a potential or indirect contribution to GHG emissions mitigation
- Y02E60/10—Energy storage using batteries
Definitions
- the present invention relates to a deterioration estimation device, a computer program, and a deterioration estimation method.
- Energy storage devices are widely used in uninterruptible power supply devices, DC or AC power supply devices included in stabilized power supplies, and the like.
- the use of power storage elements in large-scale systems that store renewable energy or power generated by existing power generation systems is expanding.
- the storage module has a configuration in which storage cells are connected in series. It is known that a power storage cell is deteriorated by repeating charge and discharge.
- Patent Document 1 discloses a secondary battery SOC (charged state) by inputting a battery voltage detected by a temperature sensor to a neural network unit that has been learned from the necessity of capacity management and safety management for a secondary battery for a vehicle. A technique for detecting the above is disclosed.
- a method such as a current integration method is used to calculate the SOC. It is possible to calculate the SOC with current data in a relatively short time. For this reason, it is easy to collect learning data for learning the neural network unit. However, since the change in SOH (State Of Health) of the power storage element takes more time than the change in SOC, estimation of SOH is not easy.
- SOH State Of Health
- An object of the present invention is to provide a degradation estimation device, a computer program, and a degradation estimation method for estimating degradation of a storage element using AI.
- a degradation estimation device that estimates degradation of a power storage element includes a SOH acquisition unit that acquires SOH at a first time point of the power storage element and SOH at a second time point after the first time point, and a second from the first time point.
- a representative value acquisition unit that acquires a representative value of the SOC of the power storage element until the time point, the SOH and the representative value at the first time point as input data, and the SOH at the second time point as output data
- a learning processing unit that learns the learning model based on the learning data.
- a computer program for causing a computer to estimate the deterioration of a power storage element includes: a process of acquiring SOH at a first time point of the power storage element and SOH at a second time point after the first time point; A process for obtaining a representative value of the SOC of the electricity storage element from the first time point to the second time point, SOH at the first time point and the representative value as input data, and SOH at the second time point as output data And a process of learning a learning model based on the learning data.
- a degradation estimation method for estimating degradation of a power storage element obtains SOH at a first time point of a power storage element and SOH at a second time point after the first time point, and between the first time point and a second time point.
- the representative value of the SOC of the storage element is acquired, and the learning model is learned based on the learning data using the SOH at the first time point and the representative value as input data and the SOH at the second time point as output data.
- the SOH acquisition unit acquires SOH at the first time point of the power storage element and SOH at the second time point after the first time point.
- the power storage element may be, for example, a power storage element operating in a moving body or facility.
- Sensor information (for example, current, voltage, and temperature) can be collected from the storage element by a monitoring device or the like.
- SOH can be estimated by a known method based on sensor information. Alternatively, SOH may be estimated using the technique described in Japanese Patent Application No. 2017-066552 (the entire contents of which are incorporated herein by reference).
- the period between the first time point and the second time point can be set as appropriate, and may be, for example, one month or three months. When the sensor information has periodicity, the period between the first time point and the second time point may be divided by the period of the sensor information. At that time, the interpolated values at the first time point and the second time point may be used as the SOH.
- the representative value acquisition unit acquires the representative value of the SOC of the storage element from the first time point to the second time point.
- a solid electrolyte interface layer SEI layer
- SEI layer solid electrolyte interface layer
- the representative value of the SOC can be said to be information that abstracts the detailed behavior of the power storage element in operation.
- the learning processing unit learns the learning model based on the learning data having the SOH and the representative value at the first time as input data and the SOH at the second time as output data.
- SOH at the first time point is represented by SOH (N)
- SOH at the second time point is represented by SOH (N + 1).
- the SOH at the first time point is SOH (N)
- the SOH at the second time point becomes SOH (N + 1) according to the representative value of the SOC of the storage element from the first time point to the second time point. Learn to be.
- Such learning data includes the SOH at the first time point, the SOH at the second time point, and the representative value of the SOC between the first time point and the second time point.
- the representative value of the SOC can be said to be a statistic obtained by abstracting or processing information for learning the learning model. Thereby, the learning model can be learned with a relatively small amount of information, and a large amount of data of the storage element is not required.
- the SOH at a certain time for example, the present
- the representative value of the SOC from the time to the prediction target time are input to the learned learning model
- the SOH at the prediction target time can be estimated. Thereby, deterioration of an electrical storage element can be estimated using AI.
- a degradation estimation device that estimates degradation of a storage element includes a SOH acquisition unit that acquires SOH at a first time point of a storage element, and a representative value of SOC of the storage element between the first time point and a second time point.
- a computer program for causing a computer to estimate deterioration of a power storage element includes: processing for acquiring SOH at a first time point of the power storage element; and processing for acquiring the power storage element between the first time point and the second time point.
- a degradation estimation method for estimating degradation of a power storage element acquires SOH at a first time point of the power storage element, acquires a representative value of SOC of the power storage element from the first time point to a second time point, and The SOH at the first time point and the representative value are input to the learned learning model to estimate the SOH at the second time point.
- the SOH acquisition unit acquires the SOH at the first time point of the power storage element.
- Sensor information for example, current, voltage, and temperature
- SOH can be collected from the storage element by a monitoring device or the like.
- SOH can be estimated by a known method based on sensor information.
- SOH may be estimated using the technique described in Japanese Patent Application No. 2017-066552 (the entire contents of which are incorporated herein by reference).
- a value estimated by a learned learning model based on the SOH at the time point before the first time point can be used. Thereby, it becomes possible to estimate SOH continuously in an arbitrary period.
- the representative value acquisition unit acquires the representative value of the SOC of the storage element from the first time point to the second time point.
- the second time point is a prediction target time point for estimating SOH.
- the representative value of the SOC of the power storage element can be obtained by abstracting the detailed behavior of the power storage element in operation from the first time point to the second time point.
- the learned learning model estimates the SOH at the second time point using the SOH and SOC representative values at the first time point as input data.
- the SOH at the first time point for example, the current time
- the representative value of the SOC of the storage element from the first time point to the second time point the prediction target time point
- the SOH at the second time point is estimated. Can do.
- how to set the representative value of the SOC of the power storage element from the first time point to the second time point determines whether the SOH at the second time point can be made larger (suppressing the decrease in SOH). You can also
- the deterioration estimation device includes an SOC estimation unit that estimates a transition of the SOC of the power storage element from the first time point to a second time point, and the representative value acquisition unit is configured to perform the first value based on the SOC transition. At least one of the total SOC fluctuation amount and the SOC fluctuation range from the time point to the second time point may be acquired.
- the SOC estimation unit estimates the transition of the SOC of the storage element from the first time point to the second time point. For example, when the time-series data amount of SOC of the electricity storage element from the first time point to the second time point is not sufficient and there is a time or time zone in which the SOC value is unknown, the obtained SOC Based on the data, the SOC transition pattern can be estimated by interpolating the SOC value where the data existed. Thereby, a large amount of sensor information is not required.
- the representative value acquisition unit acquires the average of the SOC from the first time point to the second time point based on the transition of the SOC.
- the average of the SOC is a value obtained by dividing the value obtained by sampling and summing the SOC values from the first time point to the second time point by the sampling number, and is the central SOC.
- the degree of decrease in SOH varies depending on the average of the SOC. Therefore, by learning the average SOC from the first time point to the second time point in the learning model, a learned learning model that can accurately estimate the SOH can be generated.
- the representative value acquisition unit acquires the total variation amount of the SOC from the first time point to the second time point based on the transition of the SOC.
- the total variation amount of the SOC is an integration of the variation amount along the path of how the SOC has changed from the first time point to the second time point.
- the representative value acquisition unit acquires the fluctuation range of the SOC from the first time point to the second time point based on the transition of the SOC.
- the fluctuation range of the SOC is a difference between the maximum value and the minimum value of the SOC from the first time point to the second time point.
- the degree of decrease in SOH varies depending on the variation range of the SOC. Therefore, a learned learning model capable of estimating SOH with high accuracy can be generated by causing the learning model to learn the fluctuation range of the SOC from the first time point to the second time point.
- the deterioration estimation device includes an SOC estimation unit that estimates a transition of the SOC of the power storage element between the first time point and a second time point, and the representative value acquisition unit acquires the SOC transition as the representative value. May be.
- the SOC estimation unit estimates the transition of the SOC of the storage element from the first time point to the second time point. For example, when the time-series data amount of SOC of the electricity storage element from the first time point to the second time point is not sufficient and there is a time or time zone in which the SOC value is unknown, the obtained SOC Based on the data, the SOC transition pattern can be estimated by interpolating the SOC value where the data existed. Thereby, a large amount of sensor information is not required.
- the representative value acquisition unit acquires the transition of the SOC as a representative value.
- the degree of decrease in SOH varies depending on the SOC transition pattern from the first time point to the second time point. Therefore, a learned learning model that can accurately estimate SOH can be generated by causing the learning model to learn the SOC transition pattern from the first time point to the second time point.
- the deterioration estimation device includes a temperature acquisition unit that acquires a change in temperature of the power storage element from the first time point to the second time point, and the learning processing unit learns using the temperature change as further input data.
- the learning model may be learned based on the data.
- the temperature acquisition unit acquires the transition of the temperature of the storage element from the first time point to the second time point.
- the acquired temperature transition may be an actual measurement value or an estimated value. For example, when the time-series data amount of the temperature of the storage element from the first time point to the second time point is not sufficient and there is a time or time zone in which the temperature value is unknown, the obtained temperature Based on this data, the temperature transition pattern can be estimated by interpolating the temperature at the location where the data existed. Thereby, a large amount of sensor information is not required.
- the learning processing unit learns a learning model based on learning data using temperature transition as input data.
- the degree of decrease in SOH varies depending on the temperature transition pattern from the first time point to the second time point. Therefore, a learned learning model that can estimate SOH with high accuracy can be generated by causing the learning model to learn the pattern of temperature transition from the first time point to the second time point.
- the temperature acquisition unit acquires a representative value of the temperature of the power storage element from the first time point to the second time point, and the learning processing unit obtains the representative value of the temperature as further input data.
- the learning model may be learned based on the learning data.
- the representative value of temperature is a statistic obtained by abstracting or processing information for learning a learning model.
- the learning processing unit learns a learning model based on learning data that uses a representative value of temperature as input data.
- the degree of decrease in SOH differs depending on the representative value of the temperature from the first time point to the second time point. Therefore, by learning the representative value of the temperature from the first time point to the second time point in the learning model, a learned learning model that can estimate the SOH with high accuracy can be generated.
- the learning processing unit may learn a learning model based on learning data in which an elapsed period from the time of manufacturing the power storage element to the first time is further input data.
- the learning processing unit learns a learning model based on learning data whose input data is an elapsed period from the time of manufacture of the electricity storage device to the first time.
- the SOH decreases with the standing time (calendar deterioration).
- the decrease in SOH after the first time point is considered to depend on the length of the elapsed time from the production time point (for example, the production completion time point) to the first time point. Therefore, by further generating learning data using the elapsed period from the time of manufacturing the storage element to the first time as input data, a learned learning model that can estimate SOH with high accuracy can be generated.
- the learning processing unit may learn a learning model based on learning data in which the number of times of charging / discharging from the time when the power storage element is manufactured to the first time is further input data.
- the learning processing unit is based on learning data using, as input data, the total amount of energized electricity from the time of manufacture of the power storage element to the first time (for example, the number of times of charging / discharging may be specified if the number of times of charging / discharging can be specified).
- the SOH decreases with the number of charge / discharge cycles (cycle deterioration). It is considered that the decrease in SOH after the first time point depends on the number of charge / discharge cycles from the production time point (for example, the production completion time point) to the first time point.
- the number of times of charging / discharging can be specified, by further generating learning data using as input data the number of times of charging / discharging from the time of manufacture of the electricity storage device to the first time point, a learned SOH that can be estimated accurately.
- a learning model can be generated.
- the total energized electricity amount may be used instead of the number of times of charging / discharging.
- the learning processing unit may learn a learning model based on learning data by providing a plurality of learning periods from the first time point to the second time point over a use period of the power storage element.
- the learning processing unit provides a plurality of learning periods from the first time point to the second time point over the use period of the storage element, and learns the learning model based on the learning data. Assuming that the learning period from the first time point to the second time point is ⁇ N, for example, the learning model is learned using the learning data for each of the plurality of learning periods ⁇ N over the use period from the time of manufacture of the storage element to the lifetime. By doing so, it becomes possible to estimate the SOH at a required point in time until the end of the life after manufacturing the power storage element.
- the deterioration estimation device estimates the deterioration of the storage element using the learned learning model learned by the learning processing unit.
- the SOH at the first time point for example, the current time
- the representative value of the SOC of the storage element from the first time point to the second time point the prediction target time point
- the SOH at the second time point is estimated. Can do.
- how to set the representative value of the SOC of the power storage element from the first time point to the second time point determines whether the SOH at the second time point can be made larger (suppressing the decrease in SOH). You can also
- a degradation estimation device that estimates degradation of a storage element inputs a representative value of SOC to a degradation simulator that estimates SOH of the storage element based on a change in SOC of the storage element, and obtains SOH output by the degradation simulator Learning an output value acquisition unit, an input value acquisition unit for acquiring a representative value of the SOC input to the deterioration simulator, an SOC representative value acquired by the input value acquisition unit, and an SOH acquired by the output value acquisition unit Acquired by a learning processing unit that learns a learning model using as data, a representative value acquisition unit that acquires a representative value of SOC of a storage element, an SOH acquisition unit that acquires SOH of the storage element, and the representative value acquisition unit
- the learning model learned by the learning processing unit is re-learned using the representative value of the SOC and the SOH acquired by the SOH acquisition unit as learning data. And a learning processing unit.
- a computer program for causing a computer to estimate deterioration of a power storage element is configured such that a representative value of SOC is input to a deterioration simulator that estimates SOH of the power storage element based on a change in SOC of the power storage element.
- a process for acquiring SOH output from the deterioration simulator, a process for acquiring a representative value of SOC input to the deterioration simulator, and a process for learning a learning model using the acquired representative value of SOC and the acquired SOH as learning data And a process for acquiring a representative value of the SOC of the storage element, a process for acquiring the SOH of the storage element, and the learned model using the acquired representative value of the SOC and the acquired SOH as learning data. And re-learning processing.
- a degradation estimation method for estimating degradation of a storage element inputs a representative value of SOC to a degradation simulator that estimates SOH of the storage element based on SOC variation of the storage element, and obtains SOH output by the degradation simulator
- the representative value of the SOC input to the deterioration simulator is acquired, the learning model is learned using the acquired representative value of the SOC and the acquired SOH as learning data, and the representative value of the SOC of the storage element is acquired.
- the SOH of the electricity storage element is acquired, and the learned learning model is re-learned using the acquired representative value of the SOC and the acquired SOH as learning data.
- the output value acquisition unit inputs the representative value of the SOC to the deterioration simulator that estimates the SOH of the electric storage element based on the fluctuation of the SOC of the electric storage element, and acquires the SOH output by the deterioration simulator.
- the deterioration simulator can estimate the deterioration value of the SOH of the power storage element based on the transition of the SOC based on the representative value of the SOC and the temperature of the power storage element.
- Qcnd is a non-energized deterioration value
- Qcur is an energized deterioration value.
- the coefficient K1 is a function of the SOC and the temperature T.
- the coefficient K1 is a deterioration coefficient, and the correspondence relationship between the SOC and the temperature T and the coefficient K1 may be obtained by calculation or stored in a table format.
- the SOC can be a representative value such as the center SOC, the SOC fluctuation range, and the like.
- the coefficient K2 is the same as the coefficient K1.
- the input value acquisition unit acquires the representative value of the SOC input to the deterioration simulator.
- the learning processing unit learns the learning model using the SOC representative value acquired by the input value acquisition unit and the SOH acquired by the output value acquisition unit as learning data.
- the deterioration simulator outputs a highly accurate deterioration value (that is, SOH) with respect to the input representative value of the SOC. Since the representative values of various SOCs can be generated relatively easily for simulation, learning data for the learning model can be easily generated, and the estimation accuracy (correct answer rate) of the learning model can be easily increased. .
- the representative value acquisition unit acquires the representative value of the SOC of the storage element.
- the representative value of the SOC can be referred to as information obtained by abstracting the detailed behavior of the power storage element in operation, and can be obtained from the transition of the SOC obtained from the sensor information. For example, the average value of the SOC and the total variation of the SOC Amount, variation range of SOC, etc.
- the representative value of the SOC can be obtained based on the actually measured value of the transition of the SOC, but a calculated value may be used in addition to the actually measured value.
- the SOH acquisition unit acquires the SOH of the power storage element.
- Sensor information eg, current, voltage, and temperature
- SOH can be collected from the storage element.
- SOH can be estimated by a known method based on sensor information. Alternatively, SOH may be estimated using the technique described in Japanese Patent Application No. 2017-066552 (the entire contents of which are incorporated herein by reference).
- the re-learning processing unit re-learns the learning model learned by the learning processing unit using the SOC representative value acquired by the representative value acquiring unit and the SOH acquired by the SOH acquiring unit as learning data.
- the SOC representative value or SOH obtained by collecting sensor information has a complicated transition pattern, and when these data are used as learning data to learn an unlearned learning model, the learning model is estimated. There is a tendency to require a large amount of learning data in order to improve the accuracy (correct answer rate).
- a deterioration simulator in advance, learning data can be obtained relatively easily and the estimation accuracy of an unlearned learning model can be increased. However, the accuracy of the learning model can be further increased, and the accuracy of estimating the deterioration of the storage element can be increased.
- the SOH acquisition unit acquires SOH at a first time point of the power storage element and SOH at a second time point after the first time point, and the representative value acquisition unit starts from the first time point.
- the representative value of the SOC of the electricity storage element until the second time point is acquired.
- the SOH acquisition unit acquires SOH at the first time point of the power storage element and SOH at the second time point after the first time point.
- the power storage element may be, for example, a power storage element operating in a moving body or facility.
- Sensor information (for example, current, voltage, and temperature) can be collected from the storage element by a monitoring device or the like.
- SOH can be estimated by a known method based on sensor information. Alternatively, SOH may be estimated using the technique described in Japanese Patent Application No. 2017-066552 (the entire contents of which are incorporated herein by reference).
- the period between the first time point and the second time point can be set as appropriate, and may be, for example, one month or three months. When the sensor information has periodicity, the period between the first time point and the second time point may be divided by the period of the sensor information. At that time, the interpolated values at the first time point and the second time point may be used as the SOH.
- the representative value acquisition unit acquires the representative value of the SOC of the power storage element from the first time point to the second time point.
- a solid electrolyte interface layer SEI layer
- SEI layer solid electrolyte interface layer
- the representative value of the SOC can be said to be information that abstracts the detailed behavior of the power storage element in operation, and includes, for example, the average of the SOC, the total variation of the SOC, and the fluctuation range of the SOC.
- the re-learning processing unit re-learns the learning model learned by the learning processing unit based on the learning data using the SOH and the representative value at the first time as input data and the SOH at the second time as output data.
- SOH at the first time point is represented by SOH (N)
- SOH at the second time point is represented by SOH (N + 1).
- the SOH at the first time point is SOH (N)
- the SOH at the second time point becomes SOH (N + 1) according to the representative value of the SOC of the storage element from the first time point to the second time point.
- Such learning data includes the SOH at the first time point, the SOH at the second time point, and the representative value of the SOC between the first time point and the second time point.
- the representative value of the SOC can be said to be a statistic obtained by abstracting or processing information for re-learning a learned learning model.
- the learned learning model can be relearned with a relatively small amount of information, and a large amount of data of the storage element is not required.
- the SOH at a certain time for example, the current time
- the representative value of the SOC from that time to the prediction target time are input to the learned or relearned learning model
- the SOH at the prediction target time can be estimated. .
- deterioration of an electrical storage element can be estimated using AI.
- a degradation estimation device that estimates degradation of a storage element includes a representative SOC value input to a degradation simulator that estimates SOH of the storage element based on a change in SOC of the storage element, and a representative value of the SOC to the degradation simulator.
- a learning model that learns, as learning data, SOH output from the deterioration simulator when input, the learning model further re-learns the representative value of the SOC of the storage element and the SOH of the storage element as learning data.
- a SOH acquisition unit that acquires SOH at the first time point of the power storage element; and a representative value acquisition unit that acquires a representative value of the SOC of the power storage element between the first time point and the second time point. The SOH at the first time point and the representative value acquired by the representative value acquisition unit are input to the learning model to estimate the SOH at the second time point.
- a computer program for causing a computer to estimate deterioration of a storage element is claim 18.
- a degradation estimation method for estimating degradation of a power storage element acquires SOH at a first time point of the power storage element, acquires a representative value of SOC of the power storage element from the first time point to a second time point, and The SOH and the representative value at the first time point are input to the deterioration simulator that estimates the SOH of the power storage element based on the SOC fluctuation of the power storage element, and the representative value of the SOC and the representative value of the SOC are input to the deterioration simulator.
- the SOH output from the deterioration simulator is learned as learning data
- the representative value of the SOC of the storage element and the SOH of the storage element are input to the learning model as relearning data and input to the second learning model. Estimate the SOH at the time.
- the learning model includes an SOC representative value input to the deterioration simulator that estimates the SOH of the electric storage element based on the SOC variation of the electric storage element, and an SOH output by the deterioration simulator when the representative value of the SOC is input to the deterioration simulator. Is learned as learning data.
- the representative value of the SOC of the power storage element and the SOH of the power storage element are relearned as learning data.
- the SOH acquisition unit acquires the SOH at the first time point of the power storage element.
- the representative value acquisition unit acquires the representative value of the SOC of the power storage element from the first time point to the second time point.
- the SOH at the first time point and the representative value acquired by the representative value acquisition unit are input to the learning model to estimate the SOH at the second time point.
- the SOH at a certain time for example, the current time
- the representative value of the SOC from that time to the prediction target time are input to the learned or relearned learning model
- the SOH at the prediction target time can be estimated. .
- deterioration of an electrical storage element can be estimated using AI.
- FIG. 1 is a diagram showing an outline of a remote monitoring system 100 according to the present embodiment.
- a network N including a public communication network (for example, the Internet) N1 and a carrier network N2 that implements wireless communication based on a mobile communication standard includes a thermal power generation system F, a mega solar power generation system S, wind power A power generation system W, an uninterruptible power supply (UPS) U, and a rectifier (DC power supply apparatus or AC power supply apparatus) D disposed in a stabilized power supply system for railways are connected.
- the network N is connected to a communication device 1, which will be described later, a server device 2 as a degradation estimation device that collects information from the communication device 1, and a client device 3 that acquires the collected information.
- the deterioration estimation device may be a life simulator.
- the base station BS is included in the carrier network N2, and the client device 3 can communicate with the server device 2 via the network N from the base station BS.
- an access point AP is connected to the public communication network N1, and the client device 3 can transmit and receive information to and from the server device 2 via the network N from the access point AP.
- the mega solar power generation system S, the thermal power generation system F, and the wind power generation system W are provided with a power conditioner (PCS) P and a power storage system 101.
- the power storage system 101 is configured by arranging a plurality of containers C accommodating the power storage module group L in parallel.
- the power storage module group L includes, for example, a power storage module (also referred to as a module) in which a plurality of power storage cells (also referred to as cells) are connected in series, a bank in which a plurality of power storage modules are connected in series, and a domain in which a plurality of banks are connected in parallel. It is configured with a hierarchical structure.
- the storage element is preferably a rechargeable device such as a secondary battery such as a lead storage battery and a lithium ion battery, or a capacitor. A part of the power storage element may be a primary battery that cannot be recharged.
- FIG. 2 is a block diagram showing an example of the configuration of the remote monitoring system 100.
- the remote monitoring system 100 includes a communication device 1, a server device 2, a client device 3, and the like.
- the communication device 1 is connected to the network N and is also connected to the target devices P, U, D, and M.
- the target devices P, U, D, and M include a power conditioner P, an uninterruptible power supply device U, a rectifier D, and a management device M that will be described later.
- the remote monitoring system 100 using the communication device 1 connected to each target device P, U, D, M, the state of the power storage module (power storage cell) in the power storage system 101 (for example, voltage, current, temperature, charge state ( The remote monitoring system 100 presents the detected state of the storage cell (including the deterioration state, the abnormal state, etc.) so that the user or the operator (maintenance staff) can check.
- the state of the power storage module for example, voltage, current, temperature, charge state
- the remote monitoring system 100 presents the detected state of the storage cell (including the deterioration state, the abnormal state, etc.) so that the user or the operator (maintenance staff) can check.
- the communication device 1 includes a control unit 10, a storage unit 11, a first communication unit 12, and a second communication unit 13.
- the control unit 10 includes a CPU (Central Processing Unit) and the like, and controls the entire communication device 1 using a built-in memory such as a ROM (Read Only Memory) and a RAM (Random Access Memory).
- ROM Read Only Memory
- RAM Random Access Memory
- the storage unit 11 may be a non-volatile memory such as a flash memory, for example.
- the storage unit 11 stores a device program 1P that is read and executed by the control unit 10.
- the storage unit 11 stores information collected by processing of the control unit 10 and information such as an event log.
- the first communication unit 12 is a communication interface that realizes communication with the target devices P, U, D, and M.
- a serial communication interface such as RS-232C or RS-485 can be used.
- the second communication unit 13 is an interface that realizes communication via the network N, and uses, for example, a communication interface such as Ethernet (registered trademark) or a wireless communication antenna.
- the control unit 10 can communicate with the server device 2 via the second communication unit 13.
- the client device 3 may be a computer used by an operator such as an administrator of the power storage system 101 of the power generation systems S and F and a maintenance staff of the target devices P, U, D, and M.
- the client device 3 may be a desktop or laptop personal computer, or may be a smartphone or tablet communication terminal.
- the client device 3 includes a control unit 30, a storage unit 31, a communication unit 32, a display unit 33, and an operation unit 34.
- the control unit 30 is a processor using a CPU.
- the control unit 30 causes the display unit 33 to display a web page provided by the server device 2 or the communication device 1 based on the web browser program stored in the storage unit 31.
- the storage unit 31 uses a nonvolatile memory such as a hard disk or a flash memory.
- the storage unit 31 stores various programs including a web browser program.
- the communication unit 32 uses a communication device such as a network card for wired communication, a wireless communication device for mobile communication connected to the base station BS (see FIG. 1), or a wireless communication device corresponding to connection to the access point AP. be able to.
- the control unit 30 can perform communication connection or information transmission / reception with the server device 2 or the communication device 1 via the network N by the communication unit 32.
- the display unit 33 may be a display such as a liquid crystal display or an organic EL (Electro Luminescence) display.
- the display unit 33 can display an image of a Web page provided by the server device 2 by processing based on the Web browser program of the control unit 30.
- the operation unit 34 is a user interface such as a keyboard and a pointing device that can be input and output with the control unit 30 or a voice input unit.
- the operation unit 34 may use a touch panel of the display unit 33 or a physical button provided on the housing.
- the operation unit 34 notifies the control unit 20 of operation information by the user.
- the configuration of the server device 2 will be described later.
- FIG. 3 is a diagram illustrating an example of a connection form of the communication device 1.
- the communication device 1 is connected to the management apparatus M. Further, the management apparatus M provided in each of the banks # 1 to #N is connected to the management apparatus M.
- the communication device 1 may be a terminal device (measurement monitor) that communicates with the management device M provided in each of the banks # 1 to #N and receives information on the storage element, or is connected to a power supply related device.
- a possible network card type communication device may be used.
- Each bank # 1 to #N includes a plurality of power storage modules 60, and each power storage module 60 includes a control board (CMU: Cell Monitoring Unit) 70.
- the management device M provided for each bank can communicate with the control board 70 with a communication function built in each power storage module 60 by serial communication, and the management device M connected to the communication device 1. Can send and receive information to and from.
- the management apparatus M connected to the communication device 1 aggregates information from the management apparatuses M in the banks belonging to the domain and outputs the information to the communication device 1.
- FIG. 4 is a block diagram showing an example of the configuration of the server device 2.
- the server device 2 includes a control unit 20, a communication unit 21, a storage unit 22, and a processing unit 23.
- the processing unit 23 includes an SOC estimation unit 24, an SOC representative value acquisition unit 25, a learning data generation unit 26, a learning model 27, a learning processing unit 28, and an input data generation unit 29.
- the server device 2 may be a single server computer, but is not limited to this, and may be composed of a plurality of server computers.
- the server device 2 may be a simulator.
- the control unit 20 can be constituted by a CPU, for example, and controls the entire server device 2 using a built-in memory such as a ROM and a RAM.
- the control unit 20 executes information processing based on the server program 2P stored in the storage unit 22.
- the server program 2P includes a Web server program, and the control unit 20 functions as a Web server that executes provision of a Web page to the client device 3, acceptance of login to the Web service, and the like.
- the control unit 20 can also collect information from the communication device 1 as an SNMP (Simple Network Management Protocol) server based on the server program 2P.
- SNMP Simple Network Management Protocol
- the communication unit 21 is a communication device that realizes communication connection and data transmission / reception via the network N.
- the communication unit 21 is a network card corresponding to the network N.
- the storage unit 22 may be a non-volatile memory such as a hard disk or a flash memory.
- sensor information for example, voltage data, current data, temperature data of the storage element
- states of the target devices P, U, D, and M to be monitored collected by the processing of the control unit 20 is stored.
- the processing unit 23 stores the sensor information (time-series voltage data, time-series current data, time-series temperature data) of the storage elements (storage modules, storage cells) collected in the database of the storage unit 22 for each storage element. It can be obtained by dividing into
- the processing unit 23 operates in a learning mode in which the learning model 27 is learned and an estimation mode in which an SOH (State Of Health) of the storage element is estimated (deterioration is estimated) using the learned learning model 27.
- SOH is an index that compares the current full charge capacity with reference to the full charge capacity when a new product is at a predetermined temperature. For example, 80% SOH means that only 80% of new products have capacity. Moreover, it is good also considering the electric energy which can be charged / discharged instead of a capacity
- FIG. 5 is a schematic diagram showing an example of the configuration of the learning model 27.
- the learning model 27 is a neural network model including deep learning, and includes an input layer, an output layer, and a plurality of intermediate layers.
- two intermediate layers are illustrated for convenience, but the number of intermediate layers is not limited to two and may be three or more.
- One or a plurality of nodes exist in the input layer, the output layer, and the intermediate layer, and the nodes in each layer are coupled with nodes existing in the preceding and following layers in one direction with a desired weight.
- a vector having the same number of components as the number of nodes in the input layer is given as input data (learning input data and SOH estimation input data) of the learning model 27.
- the input data includes power storage element information, time information, SOC average, total SOC variation, SOC variation, temperature average, SOH information, and the like.
- the output data includes SOH information. Details of these information will be described later. In particular, information related to SOC and temperature information are related to each other, and local fluctuations in information related to SOC and temperature information are related to changes in SOH. It is desirable to use a convolutional neural network added to the above.
- the output of the intermediate layer is calculated using the weight and activation function, and the calculated value is transferred to the next intermediate layer. In the same manner, the output is successively transmitted to subsequent layers (lower layers) until the output of the output layer is obtained. Note that all of the weights for joining the nodes are calculated by a learning algorithm.
- the output layer of the learning model 27 generates SOH information as output data.
- the output data may be vector data having a component having the same size as the number of nodes in the output layer (size of the output layer). For example, when the SOH of the storage element is output every 1% from 0% to 100%, the number of nodes in the output layer can be set to 101. As practical SOH information, for example, when SOH is estimated every 1% between 60% and 100%, the number of nodes in the output layer may be 41. Alternatively, the SOH value may be divided every several percent, and output data may be output every several percent. The output value from the output layer can be interpreted as the probability of being classified into each category of SOH.
- the learning model 27 and the learning processing unit 28 include, for example, a CPU (for example, a multiprocessor equipped with a plurality of processor cores), a GPU (Graphics Processing Units), a DSP (Digital Signal Processors), an FPGA (Field-Programmable Gate Arrays). ) And the like can be combined. A quantum processor can also be combined.
- the learning model 27 is not limited to the neural network model, and may be another machine learning model.
- FIG. 6 is a schematic diagram showing an example of a decrease in SOH according to the usage time of the power storage element.
- the vertical axis represents SOH (%)
- the horizontal axis represents time.
- the SOH decreases with the use time (including the leaving time).
- time points ta, tb, tc, and td are set, and the time points between time point tb and time point ta are the same as the time points between time point td and time point tc.
- the SOH decrease ⁇ SOH (tb) from the time point ta to the time point tb is different from the SOH decrease ⁇ SOH (td) from the time point tc to the time point td.
- the degree of decrease in SOH varies depending on the usage state of the power storage element. Therefore, in order to identify various usage states of the storage element, grasping the usage state of the storage element between two different time points is an important factor for estimating the SOH of the storage element.
- FIG. 7 is a schematic diagram showing an example of the relationship between temperature and SOH reduction in a standing test at a constant SOC.
- the vertical axis represents SOH
- the horizontal axis represents time.
- the degree of decrease in SOH is larger as the temperature of the power storage element is higher. That is, there is a correlation between a decrease (decrease) in SOH and a temperature (average temperature), and temperature (average temperature) is important information when estimating (predicting) SOH.
- FIG. 8 is a schematic diagram showing an example of the relationship between the average SOC and SOH reduction in a standing test at a constant temperature.
- the vertical axis represents SOH
- the horizontal axis represents time.
- the average of the SOC is a value obtained by dividing a value obtained by sampling and summing the SOC values within a certain period by the sampling number, and is a central SOC.
- FIG. 8 it can be seen that the degree of decrease in SOH is different when the average SOC is different. That is, there is a correlation between the decrease (decrease) in SOH and the average SOC, and the average SOC is an important representative value when estimating (predicting) SOH.
- FIG. 9 is a schematic diagram showing an example of the relationship between the SOC fluctuation range and the SOH decrease in the cycle test at a constant temperature and a constant SOC average.
- the vertical axis represents SOH
- the horizontal axis represents time.
- the fluctuation range of the SOC is a difference between the maximum value and the minimum value of the SOC in the period when the SOC fluctuates within a certain period.
- the greater the variation range of the SOC the greater the degree of decrease in SOH. That is, there is a correlation between the decrease (decrease) in SOH and the fluctuation range of the SOC, and the fluctuation width of the SOC becomes an important representative value when estimating (predicting) the SOH.
- FIG. 10 is a schematic diagram showing an example of the relationship between the total SOC fluctuation amount and the SOH reduction in the cycle test at a constant temperature and a constant SOC fluctuation range.
- the vertical axis represents SOH
- the horizontal axis represents time.
- the SOC total fluctuation amount is an integration of fluctuation amounts along the path of how the SOC fluctuates in a certain period.
- the degree of decrease in SOH increases as the total SOC fluctuation amount increases. That is, there is a correlation between the decrease (decrease) in SOH and the total SOC variation, and the total SOC variation is an important representative value when estimating (predicting) SOH.
- FIG. 11 is a schematic diagram illustrating an example of a decrease in SOH of the storage element at the first time point and the second time point.
- SOH at the first time point is represented by SOH (N)
- SOH at the second time point is represented by SOH (N + 1).
- the decrease in SOH at the second time point ⁇ SOH (N) ⁇ SOH (N + 1) ⁇ is not limited to the SOH (N) at the first time point, but what the power storage element does from the first time point to the second time point. It depends on whether it was in use.
- the use state of the power storage element from the first time point to the second time point is known by specifying the SOH (N) at the first time point of the power storage element, the SOH at the second time point of the power storage element is obtained. (N + 1) can be estimated. Further, as described above, the usage state from the first time point to the second time point can be characterized by the temperature (average temperature) of the power storage element and the representative value related to the SOC.
- the processing unit 23 has a function as an SOH acquisition unit, and acquires the SOH at the first time point of the power storage element and the SOH at the second time point after the first time point.
- the power storage element can be, for example, a power storage element that is actually operating in a moving object or facility.
- SOH can be estimated by a known method based on sensor information. Alternatively, SOH may be estimated using the technique described in Japanese Patent Application No. 2017-066552 (the entire contents of which are incorporated herein by reference).
- the period between the first time point and the second time point can be set as appropriate, and may be, for example, one month or three months. When the sensor information has periodicity, the period between the first time point and the second time point may be divided by the period of the sensor information.
- the interpolated values at the first time point and the second time point may be used as the SOH. Further, as the SOH at the first time point, the value estimated by the learned learning model 27 based on the SOH at the time point before the first time point can be used. This makes it possible to continuously estimate the SOH of the power storage element in an arbitrary period during the use period (including the leaving period) after the power storage element is manufactured.
- the SOC representative value acquisition unit 25 has a function as a representative value acquisition unit, and acquires the representative value of the SOC of the storage element from the first time point to the second time point.
- a storage element for example, a lithium ion battery
- SEI layer solid electrolyte interface layer
- the inventor of the present application has come up with the idea of acquiring a representative value of the SOC that affects the decrease in SOH of the power storage element and using it for learning of the AI.
- the representative value of the SOC can be said to be information that abstracts the detailed behavior of the power storage element in operation. Note that the representative value of the SOC may be acquired from an external server device or the like.
- the learning data generation unit 26 generates learning data using the SOH at the first time point and the SOC representative value between the first time point and the second time point as input data, and the SOH at the second time point as output data. .
- the learning processing unit 28 learns the learning model 27 based on the learning data generated by the learning data generating unit 26.
- the above-described learning data generation unit 26 does not need to be included in the server device 2 but is included in another server device, acquires learning data generated by the server device, and the learning processing unit 28 Alternatively, the learning model 27 may be learned based on the acquired learning data. The same applies to the following description of this specification.
- SOH at the first time point is represented by SOH (N)
- SOH at the second time point is represented by SOH (N + 1).
- the SOH at the first time point is SOH (N)
- the SOH at the second time point is SOH (N + 1) according to the representative value of the SOC of the storage element from the first time point to the second time point. Learn to become.
- Such learning data includes the SOH at the first time point, the SOH at the second time point, and the representative value of the SOC between the first time point and the second time point.
- the representative value of the SOC can be said to be a statistic obtained by abstracting or processing information for learning the learning model 27.
- the learning model 27 can be learned with a relatively small amount of information, and a large amount of data of the storage element is not required.
- the SOH at a certain time for example, the present
- the representative value of the SOC from the time to the prediction target time are input to the learned learning model 27, the SOH at the prediction target time can be estimated. Thereby, deterioration of an electrical storage element can be estimated using AI.
- FIG. 12 is a schematic diagram showing an example of temperature data from the first time point to the second time point.
- the vertical axis represents temperature and the horizontal axis represents time.
- the processing unit 23 calculates an average temperature between the first time point and the second time point based on the temperature data between the first time point and the second time point.
- the temperature data is obtained in detail between the first time point and the second time point, but even if the number of data is small, the average temperature between the first time point and the second time point Can be requested.
- FIG. 13 is a schematic diagram showing an example of the SOC transition pattern from the first time point to the second time point.
- the vertical axis represents SOC and the horizontal axis represents time.
- the SOC estimation unit 24 estimates the transition of the SOC of the storage element from the first time point to the second time point.
- the time-series data of the SOC of the power storage element from the first time point to the second time point is obtained in detail, but even when the number of data is small, the first time point and the second time point data are obtained.
- a pattern of SOC transition between the time points can be obtained.
- the obtained SOC Based on the data the SOC transition pattern can be estimated by interpolating the SOC value where the data existed. Thereby, a large amount of sensor information is not required.
- FIG. 14 is a schematic diagram showing an example of a representative value of SOC between the first time point and the second time point.
- the vertical axis represents SOC and the horizontal axis represents time.
- the SOC representative value acquisition unit 25 acquires the average of the SOC from the first time point to the second time point based on the SOC transition pattern.
- the average of the SOC is a value obtained by dividing the value obtained by sampling and summing the SOC values from the first time point to the second time point by the sampling number, and is the central SOC.
- the degree of decrease in SOH varies depending on the average of the SOC.
- the learning data generation unit 26 generates learning data using the average of the SOC from the first time point to the second time point as input data. By causing the learning model 27 to learn the average SOC from the first time point to the second time point, the learned learning model 27 that can accurately estimate the SOH can be generated.
- the SOC representative value acquisition unit 25 acquires the total amount of SOC variation from the first time point to the second time point based on the SOC transition pattern.
- the total variation amount of the SOC is an integration of the variation amount along the path of how the SOC has changed from the first time point to the second time point.
- the learning data generation unit 26 generates learning data using as input data the total amount of SOC variation from the first time point to the second time point. By causing the learning model 27 to learn the total SOC variation from the first time point to the second time point, it is possible to generate the learned learning model 27 that can accurately estimate the SOH.
- the SOC representative value acquisition unit 25 acquires the fluctuation range of the SOC from the first time point to the second time point based on the SOC transition pattern.
- the fluctuation range of the SOC is a difference between the maximum value and the minimum value of the SOC from the first time point to the second time point.
- the learning data generation unit 26 generates learning data with the variation range of the SOC from the first time point to the second time point as input data. By causing the learning model 27 to learn the SOC fluctuation range from the first time point to the second time point, it is possible to generate the learned learning model 27 that can accurately estimate the SOH.
- the representative value of the SOC is the average of the SOC from the first time point to the second time point, the total variation amount of the SOC from the first time point to the second time point, The variation range of the SOC from the first time point to the second time point can be set.
- the learning data generation unit 26 generates learning data using the average temperature between the first time point and the second time point as input data.
- the average temperature can also be referred to as a transition of temperature between the first time point and the second time point.
- FIG. 15 is a block diagram showing an example of learning data.
- the time width from the first time point to the second time point is represented by ⁇ t.
- the learning input data includes the time width ⁇ t from the first time point to the second time point, the average temperature in the time width ⁇ t, the average of the SOC in the time width ⁇ t, and the SOC in the time width ⁇ t.
- the total fluctuation amount, the fluctuation width of the SOC in the time width ⁇ t, and the SOH (N) at the first time point can be used.
- the learning output data can be SOH (N + 1) at the second time point. Note that the learning input data does not need to include all of the data shown in FIG. 15. For example, either or both of the total variation amount of SOC and the variation range of SOC may be omitted. However, it is expected to improve the estimation accuracy of SOH by learning the total variation amount of SOC and the variation range of SOC.
- FIG. 16 is a block diagram showing another example of learning data.
- the learning input data further includes an elapsed time Dt1 from the production time (for example, production completion time) to the first time, and the production time (for example, production completion time) to the first time point.
- the number of charge / discharge cycles Ct1 can be included.
- the learning data generation unit 26 can generate learning data whose input data is an elapsed period from the time when the power storage element is manufactured to the first time.
- the SOH decreases with the standing time (calendar deterioration).
- the decrease in SOH after the first time point is considered to depend on the length of the elapsed time from the production time point (for example, the production completion time point) to the first time point. Therefore, by further generating learning data using the elapsed period from the time of manufacturing the storage element to the first time as input data, a learned learning model 27 that can estimate SOH with high accuracy can be generated.
- the learning data generation unit 26 uses as input data the total amount of energized electricity from the time of manufacture of the storage element to the first time (for example, the number of times of charging / discharging may be specified if the number of times of charging / discharging can be specified). Can be generated.
- the SOH decreases with the number of charge / discharge cycles (cycle deterioration). It is considered that the decrease in SOH after the first time point depends on the number of charge / discharge cycles from the production time point (for example, the production completion time point) to the first time point.
- the number of times of charging / discharging can be specified, by further generating learning data using as input data the number of times of charging / discharging from the time of manufacture of the electricity storage device to the first time point, a learned SOH that can be estimated accurately.
- a learning model 27 can be generated. Note that the learning data may be generated using only one of the elapsed time Dt1 and the charge / discharge count Ct1. In addition, when the number of times of charging / discharging cannot be specified, the total amount of energized electricity may be used instead of the number of times of charging / discharging.
- FIG. 17 is a schematic diagram showing an example in which a plurality of learning periods are set.
- the vertical axis represents SOH
- the horizontal axis represents time.
- time point N1, time point N2, time point N3, time point N4, time point N5, time point N6, time point N7, and time point N8 are set.
- the time between adjacent time points may be different.
- a learning period in which the time point N1 is the first time point and the time point N2 is the second time point can be set.
- a learning period in which the time point N2 is the first time point and the time point N3 is the second time point can be set. The same applies hereinafter.
- the learning data generation unit 26 can generate learning data by providing a plurality of learning periods from the first time point to the second time point over the use period of the storage element. Assuming that the learning period from the first time point to the second time point is ⁇ N, for example, the learning model 27 is obtained using the learning data of each of the plurality of learning periods ⁇ N over the use period from the time of manufacture of the power storage element to the lifetime. By learning, it becomes possible to estimate the SOH at a required point in time until the end of the life after manufacturing the power storage element. That is, it is possible to predict the life of the power storage element.
- FIG. 18 is a block diagram showing an example of input data for life estimation.
- the input data for life estimation includes the time width ⁇ te of the prediction period, the average temperature in the time width ⁇ te, the average of the SOC in the time width ⁇ te, and the total fluctuation amount of the SOC in the time width ⁇ te.
- the time width ⁇ te can be the same as the time width from the first time point to the second time point used in the learning data. Note that the input data for life estimation need not include all of the data shown in FIG.
- either the total variation amount of SOC, the variation range of SOC, or both may be omitted.
- the total SOC fluctuation amount and SOC fluctuation range may be added to the input data, it is possible to expect improvement in the estimation accuracy of SOH.
- the input data generation unit 29 generates input data for life estimation based on the SOH at the first time point and the representative value of the SOC of the storage element from the first time point to the second time point. Specifically, the input data generation unit 29, as illustrated in FIG. 18, calculates the time width ⁇ te of the prediction period, the average temperature in the time width ⁇ te, the average of the SOC in the time width ⁇ te, and the SOC in the time width ⁇ te. Input data related to the total fluctuation amount, the fluctuation width of the SOC in the time width ⁇ te, and the start point (first time point) SOHe of the prediction period. For example, as the SOHe at the first time point, a value estimated by the learned learning model 27 based on the SOH at the time point before the first time point can be used. Thereby, it becomes possible to estimate SOH continuously in an arbitrary period.
- the learned learning model 27 estimates the SOH at the prediction target time point (the end point of the prediction period, that is, the second time point) based on the input data generated by the input data generation unit 29. Accordingly, if the SOH at the first time point (for example, the current time) and the representative value of the SOC of the power storage element in the prediction period are known, the SOH at the prediction target time point can be estimated. In addition, by using the SOH at the first time point, how to set the representative value of the SOC of the power storage element in the prediction period with the first time point as a reference, the SOH at the prediction target time point becomes larger (SOH It is also possible to determine whether or not a decrease in the amount can be suppressed.
- FIG. 19 is a block diagram showing another example of input data for life estimation.
- the input data further includes an elapsed time Dte from the production time (for example, production completion time) to the start point of the prediction period, and the start time of the prediction period from the production time (for example, production completion time).
- the number of charging / discharging times Cte is included.
- FIG. 20 is a schematic diagram showing an example of the degradation estimation result of the storage element.
- the vertical axis represents SOH
- the horizontal axis represents time.
- the learned learning model 27 estimates the SOH at the future prediction target time at the start point (for example, present) of the prediction period.
- a curved line indicated by a broken line shows a transition of SOH in the planned use condition (for example, the use state scheduled by the user) from the current storage element to the prediction target time.
- the SOH can be estimated by changing the use conditions from the current storage element to the prediction target time.
- a use condition (recommended use condition) that can suppress a decrease in SOH at the prediction target time point as a curve indicated by a solid line.
- SOH current (starting point) SOH
- the SOH can be reduced. It is possible to provide the user with information on whether or not it can be suppressed.
- FIG. 21 is a flowchart showing an example of a processing procedure of the processing unit 23 in the learning mode.
- the processing unit 23 sets the first time point and the second time point (S11), and acquires the SOH at the first time point and the second time point (S12).
- the processing unit 23 acquires sensor information (time series current data, time series temperature data, time series voltage data) from the first time point to the second time point (S13).
- the processing unit 23 generates an SOC transition pattern in the time width ⁇ t from the first time point to the second time point based on the acquired sensor information (S14), and acquires a representative value of the SOC in the time width ⁇ t (S15). ).
- the processing unit 23 generates learning data using the time width ⁇ t, the representative value of the SOC, the average temperature in the time width ⁇ t, and the SOH at the first time point as input data, and the SOH at the second time point as output data ( S16).
- the processing unit 23 learns and updates the learning model 27 based on the generated learning data (S17), and determines whether or not to end the processing (S18). When it is determined that the process is not to be ended (NO in S18), the processing unit 23 continues the process after step S12, and when it is determined that the process is to be ended (YES in S18), the process is ended.
- FIG. 22 is a flowchart illustrating an example of a processing procedure of the processing unit 23 in the estimation mode.
- the processing unit 23 sets a prediction period (S31), and acquires SOH at the start point (first time point) of the prediction period (S32).
- the processing unit 23 acquires sensor information for the prediction period (S33), and generates an SOC transition pattern for the prediction period based on the acquired sensor information (S34).
- the processing unit 23 acquires the representative value of the SOC in the prediction period (S35), and generates prediction input data based on the time width of the prediction period, the representative value of the SOC, the average temperature, and the SOH at the start point of the prediction period. (S36). The processing unit 23 estimates SOH at the prediction target time (S37) and ends the process.
- the server device 2 of the present embodiment a large amount of data is not required when the learning model 27 is learned. Since the representative value for the SOC that characterizes the usage state of the storage element can be acquired and the learning model 27 can be learned based on the acquired representative value, it is possible to learn even with a small amount of data. In particular, it is possible to predict the deterioration of the storage element even at a stage where it is difficult to obtain a large amount of sensor information during actual operation of the storage element, such as the development phase of the storage element or the initial stage of the manufacturing phase. Further, since the sensor information is not used as it is as learning data, the influence of errors included in the sensor information can be suppressed.
- the learning model 27 may use a convolutional neural network (CNN) in which a convolution layer or a pooling layer is added to the intermediate layer. Thereby, it is possible to learn by reflecting the relationship between the SOC information and the temperature information, and the relationship between the local variation of the SOC information and the temperature information and the change of the SOH.
- CNN convolutional neural network
- the abstract data called the representative value of the SOC is used.
- the present invention is not limited to this.
- learning data can be generated in a manner closer to time-series data such as sensor information by reducing the degree of abstraction.
- FIG. 23 is a schematic diagram showing an example of the SOC transition pattern of the storage element from the first time point to the second time point.
- the SOC transition pattern shown in FIG. 23 is the same as that shown in FIG.
- the SOC estimation unit 24 generates an SOC transition pattern of the storage element from the first time point to the second time point. For example, when the time-series data amount of SOC of the electricity storage element from the first time point to the second time point is not sufficient and there is a time or time zone in which the SOC value is unknown, the obtained SOC Based on the data, the SOC transition pattern can be generated by interpolating the SOC value at the location where the data existed. Thereby, a large amount of sensor information is not required.
- the SOC representative value acquisition unit 25 acquires the SOC transition pattern generated by the SOC estimation unit 24 as a representative value.
- the degree of decrease in SOH varies depending on the SOC transition pattern from the first time point to the second time point.
- the SOC representative value can be a transition pattern of the SOC from the first time point to the second time point.
- FIG. 24 is a schematic diagram showing an example of a temperature transition pattern of the storage element from the first time point to the second time point.
- the vertical axis represents temperature and the horizontal axis represents time.
- the processing unit 23 generates a temperature transition pattern of the storage element from the first time point to the second time point. For example, when the time-series data amount of the temperature of the storage element from the first time point to the second time point is not sufficient and there is a time or time zone in which the temperature value is unknown, the obtained temperature Based on this data, the temperature transition pattern can be estimated by interpolating the temperature at the location where the data existed. Thereby, a large amount of sensor information is not required.
- the processing unit 23 has a function as an acquisition unit that acquires a temperature transition, and may acquire a temperature transition pattern, or may acquire an actual measurement value of a temperature transition of the storage element. Further, the processing unit 23 may acquire a representative value of temperature.
- the representative value of the temperature can be a statistic obtained by abstracting or processing information for learning the learning model 27. For example, the representative value of the temperature is the same as the representative value of the SOC shown in FIG.
- the learning data generation unit 26 can generate learning data having the SOC transition pattern and the temperature transition pattern as input data.
- the degree of decrease in SOH varies depending on the temperature transition pattern from the first time point to the second time point. Therefore, the learned learning model 27 that can estimate the SOH with high accuracy can be generated by causing the learning model 27 to learn the SOC transition pattern and the temperature transition pattern from the first time point to the second time point.
- the first time point and the second time point may be divided by a plurality of time widths, and values for each time width of the time series of the SOC transition pattern and the temperature transition pattern may be used.
- the learning data is similar to the first embodiment in that the SOH at the first time point is input data and the SOH at the second time point is output data.
- the learning data generating unit 26 can generate learning data using the SOC transition pattern and the measured value or representative value of the temperature transition as input data.
- the second embodiment at a stage where more sensor information can be collected than in the case of the first embodiment, it is possible to predict deterioration reflecting the characteristics of the sensor information itself.
- the learning processing unit 28 (processing unit 23) is configured to learn and update the learning model 27.
- the learning processing unit 28 (processing The unit 23) learns the learning model 27 using the deterioration value (SOH) of the storage element output from the deterioration simulator described later as learning data, and learns using the same processing as in the first and second embodiments.
- the learned learning model 27 can be re-learned.
- the third embodiment will be described.
- FIG. 25 is a schematic diagram showing the operation of the deterioration simulator 50 of the third embodiment.
- Degradation simulator 50 estimates (calculates) a degradation value of the storage element when acquiring the SOC pattern and the temperature pattern as input data.
- the SOC pattern can be specified by the representative value of the SOC.
- the representative value of the SOC affects the deterioration of the power storage element, and includes, for example, the SOC average (also referred to as the central SOC), the SOC fluctuation range, the total fluctuation amount of the SOC, and the like.
- the deterioration simulator 50 can estimate the deterioration value of the SOH of the power storage element based on the transition of the SOC based on the representative value of the SOC and the temperature of the power storage element. If SOH at time t (also referred to as health) is SOH t, and SOH at time t + 1 is SOH t + 1 , the degradation value is (SOH t ⁇ SOH t + 1 ). That is, if the SOH at the time t is known, the SOH at the time t + 1 can be obtained based on the deterioration value.
- the time point can be a current time point or a future time point
- the time point t + 1 can be a time point when a required time has elapsed from the time point t toward the future.
- the time difference between the time point t and the time point t + 1 is a deterioration prediction target period of the deterioration simulator 50, and can be set as appropriate according to how much future the deterioration value is predicted.
- the time difference between the time point t and the time point t + 1 can be, for example, a required time such as one month, six months, one year, or two years.
- a temperature pattern is input, but a required temperature (for example, an average temperature from time t to time t + 1) may be input instead of the temperature pattern.
- Qcnd is a non-energized deterioration value
- Qcur is an energized deterioration value.
- the coefficient K1 is a function of the SOC and the temperature T.
- t is an elapsed time, for example, a time from time t to time t + 1.
- the coefficient K1 is a deterioration coefficient, and the correspondence relationship between the SOC and the temperature T and the coefficient K1 may be obtained by calculation or stored in a table format.
- the SOC can be a representative value such as the center SOC, the SOC fluctuation range, and the like.
- the coefficient K2 is the same as the coefficient K1.
- FIG. 26 is a schematic diagram showing an example of SOH estimation by the deterioration simulator 50.
- SOC patterns representative values of three SOCs indicated by symbols A, B, and C are input to the deterioration simulator 50.
- the SOC pattern indicated by symbol A has a relatively small fluctuation range of the SOC.
- the SOC pattern indicated by symbol C has a relatively large variation range of the SOC.
- the SOC pattern indicated by reference character B the fluctuation range of the SOC is approximately between the two.
- the estimation of SOH by the deterioration simulator 50 is indicated by symbols A, B, and C.
- the SOH transition of the code A corresponds to the SOC pattern of the code A
- the SOH transition of the code B corresponds to the SOC pattern of the code B
- the SOH transition of the code C corresponds to the SOC pattern of the code C. ing. In this way, it can be seen that the deterioration value tends to increase as the fluctuation range of the SOC increases.
- FIG. 26 three comparatively simple SOC patterns are illustrated for convenience, but by inputting various SOC patterns to the deterioration simulator 50, deterioration values (SOH) corresponding to various SOC patterns may be estimated. In addition, it is possible to easily generate a large amount of data sets of SOH estimation results corresponding to the SOC pattern.
- SOH deterioration values
- the processing unit 23 has a function as an input value acquisition unit, and can acquire the representative value of the SOC input to the deterioration simulator 50. Further, the processing unit 23 has a function as an output value acquisition unit, and can acquire the SOH output by the deterioration simulator 50 when the representative value of the SOC is input to the deterioration simulator 50. That is, the processing unit 23 acquires the representative value of the SOC input to the deterioration simulator 50 and the SOH (obtained from the deterioration value) output by the deterioration simulator 50 at that time.
- the learning processing unit 28 learns the learning model 27 using the acquired representative value of the SOC and the acquired SOH as learning data.
- FIG. 27 is a block diagram showing an example of learning data based on the estimated value of the deterioration simulator 50.
- the time width from time t to time t + 1 is represented by ⁇ t.
- the learning input data includes a time width ⁇ t from time t to time t + 1, an average temperature in time width ⁇ t, an average of SOC in time width ⁇ t, and a total variation of SOC in time width ⁇ t.
- the amount of SOC, the fluctuation range of the SOC at the time width ⁇ t, and the SOH (N) at the time point t can be used.
- the learning output data can be SOH (N + 1) at time t + 1.
- the learning input data does not need to include all of the data shown in FIG. 27.
- either or both of the total variation amount of SOC and the variation range of SOC may be omitted.
- the deterioration simulator 50 When representative values of various patterns of SOC are input to the deterioration simulator 50, the deterioration simulator 50 outputs a highly accurate deterioration value (that is, SOH) with respect to the input representative value of the SOC. Can be easily generated, and the estimation accuracy (correct answer rate) of the learning model 27 can be easily increased.
- SOH highly accurate deterioration value
- the relearning process by the learning processing unit 28 is the same as the learning process in the first embodiment and the second embodiment.
- the SOC representative value acquisition unit 25 has a function as a representative value acquisition unit, and acquires a representative value of the SOC of the storage element.
- the representative value of the SOC can be referred to as information obtained by abstracting the detailed behavior of the power storage element in operation, and can be obtained from the transition of the SOC obtained from the sensor information. For example, the average value of the SOC and the total variation of the SOC Amount, variation range of SOC, etc.
- the representative value of the SOC can be obtained based on the actually measured value of the transition of the SOC, but a calculated value may be used in addition to the actually measured value.
- the SOC representative value acquisition unit 25 acquires the representative value of the SOC of the power storage element from the first time point to the second time point.
- a solid electrolyte interface layer SEI layer
- SEI layer solid electrolyte interface layer
- the inventor of the present application has come up with the idea of acquiring a representative value of the SOC that affects the decrease in SOH of the power storage element and using it for learning of the AI.
- the representative value of the SOC can be said to be information that abstracts the detailed behavior of the power storage element in operation, and includes, for example, the average of the SOC, the total variation of the SOC, and the fluctuation range of the SOC.
- the processing unit 23 has a function as an SOH acquisition unit, and acquires the SOH of the power storage element.
- Sensor information eg, current, voltage, and temperature
- SOH can be collected from the storage element.
- SOH can be estimated by a known method based on sensor information. Alternatively, SOH may be estimated using the technique described in Japanese Patent Application No. 2017-066552 (the entire contents of which are incorporated herein by reference).
- the processing unit 23 acquires the SOH at the first time point of the power storage element and the SOH at the second time point after the first time point.
- the power storage element may be, for example, a power storage element operating in a moving body or facility.
- Sensor information (for example, current, voltage, and temperature) can be collected from the storage element by a monitoring device or the like.
- SOH can be estimated by a known method based on sensor information. Alternatively, SOH may be estimated using the technique described in Japanese Patent Application No. 2017-066552 (the entire contents of which are incorporated herein by reference).
- the period between the first time point and the second time point can be set as appropriate, and may be, for example, one month or three months.
- the sensor information has periodicity, the period between the first time point and the second time point may be divided by the period of the sensor information. At that time, the interpolated values at the first time point and the second time point may be used as the SOH.
- the learning processing unit 28 has a function as a re-learning processing unit, and re-learns the learning model 27 using the acquired representative value of the SOC and the acquired SOH as learning data.
- the learning processing unit 28 re-reads the learned learning model 27 based on the learning data having the SOH and the representative value at the first time as input data and the SOH at the second time as output data. Let them learn.
- SOH at the first time point is represented by SOH (N)
- SOH at the second time point is represented by SOH (N + 1).
- the SOH at the first time point is SOH (N)
- the SOH at the second time point becomes SOH (N + 1) according to the representative value of the SOC of the storage element from the first time point to the second time point.
- Such learning data includes the SOH at the first time point, the SOH at the second time point, and the representative value of the SOC between the first time point and the second time point.
- the representative value of the SOC can be said to be a statistic obtained by abstracting or processing information for re-learning the learned learning model 27.
- the learned learning model can be relearned with a relatively small amount of information, and a large amount of data of the storage element is not required.
- the SOH at a certain time for example, the current time
- the representative value of the SOC from that time to the prediction target time are input to the learned or re-learned learning model 27, the SOH at the prediction target time can be estimated. It can. Thereby, deterioration of an electrical storage element can be estimated using AI.
- FIG. 28 is a schematic diagram showing an example of the relationship between the learning data set of the learning model 27 and the estimation accuracy.
- the vertical axis represents the estimation accuracy (correct answer rate) of the learning model 27
- the horizontal axis represents the learning data set (data amount required for learning).
- a curve indicated by a symbol R ′ indicates a case of learning based only on actual measurement values (for example, market data).
- the SOC representative value and SOH obtained by collecting sensor information, such as the curve indicated by the symbol R ′ have a complicated transition pattern, for example, and these data are used as learning data for unlearned learning.
- the model 27 is learned, there is a tendency that a large amount of learning data is required to increase the estimation accuracy (correct answer rate) of the learning model 27.
- a curve indicated by a symbol Sa represents a case where learning is performed using the data of the deterioration simulator 50 as learning data.
- the deterioration simulator 50 outputs a highly accurate deterioration value (that is, SOH) with respect to the input representative value of the SOC.
- SOH highly accurate deterioration value
- the estimation accuracy (correct answer rate) of the learning model 27 can be easily increased.
- a deterioration simulator in advance indicated by a curve indicated by symbol Sa
- learning data can be obtained relatively easily and the estimation accuracy of an unlearned learning model can be increased.
- the parameters of the deterioration simulator 50 are determined as shown by the curve of the symbol Sb, there is a tendency that a large amount of data is difficult to reflect, and there is a tendency that the estimation accuracy is difficult to improve for the amount of data.
- a curve indicated by a symbol R represents a case where re-learning with an actual measurement value is performed after learning using the estimated value of the deterioration simulator 50. Like the curve indicated by the symbol R, the accuracy of the learning model 27 can be further increased without a large amount of learning data, and the accuracy of the deterioration estimation of the storage element can be increased.
- the SOC representative value input to the deterioration simulator 50 and the SOH output by the deterioration simulator 50 when the SOC representative value is input to the deterioration simulator 50 are learned as learning data.
- the learning model 27 further re-learns the representative value of the SOC of the power storage element and the SOH of the power storage element as learning data.
- the server apparatus 2 acquires the SOH at the first time point of the power storage element.
- the server device 2 acquires a representative value of the SOC of the power storage element from the first time point to the second time point.
- the server device 2 can input the SOH at the first time point and the acquired representative value of the SOC to the learning model 27 to estimate the SOH at the second time point.
- the server device 2 will have the SOH at the prediction target time point. Can be estimated. Thereby, deterioration of an electrical storage element can be estimated using AI.
- FIG. 29 is a flowchart illustrating an example of a processing procedure of the processing unit 23 in the learning mode of the third embodiment.
- the processing unit 23 inputs the representative SOC value and temperature pattern to the deterioration simulator (S41), acquires the SOH (deterioration value) output by the deterioration simulator (S42), and represents the representative SOC value input to the deterioration simulator 50, A temperature pattern is acquired (S43).
- the process of step S41 can be previously performed with another apparatus, without performing with the process part 23.
- the processing unit 23 generates learning data using the SOC representative value, temperature pattern, and SOH at time t as input data, and SOH at time t + 1 as output data (S44), and using the generated learning data.
- the learning model 27 is learned (S45).
- the processing unit 23 determines the presence / absence of data of the deterioration simulator 50 (S46), and if there is data (YES in S46), the processing after step S41 is continued. When there is no data (NO in S46), the processing unit 23 re-learns the learning model 27 using the measured value data (S47), and ends the process.
- the process of step S47 can be the same process as the process illustrated in FIG.
- the server apparatus 2 has the configuration including the learning model 27 and the learning processing unit 28, but is not limited thereto.
- the learning model 27 and the learning processing unit 28 may be provided in another one or a plurality of servers.
- the deterioration estimation device is not limited to the server device 2.
- an apparatus such as a life estimation simulator may be used.
- the deterioration simulator 50 can be incorporated into the server device 2, but can also be configured as a device separate from the server device 2.
- the control unit 20 and the processing unit 23 of the present embodiment can be realized using a general-purpose computer including a CPU (processor), a GPU, a RAM (memory), and the like. That is, as shown in FIGS. 21, 22, and 29, a computer program that defines the procedure of each process is loaded into a RAM (memory) provided in the computer, and the computer program is executed by a CPU (processor).
- the control unit 20 and the processing unit 23 can be realized on the computer.
- the computer program may be recorded on a recording medium and distributed.
- the learning model 27 learned by the server device 2 and the computer program based on the learning model 27 may be distributed and installed to the remote monitoring target devices P, U, D, M and terminal devices via the network N and the communication device 1.
- the learning model 27 may be learned using the representative value in the initial stage of learning, and then learned using the actual measurement value.
- the data amount of the representative value and the actual measurement value may be changed as appropriate according to the learning stage, or the ratio between the two may be changed.
- the learning data can be divided into so-called training data and test data, and the learning model 27 is first learned using the training data, and then the learning model 27 can be evaluated using the test data.
- server device 20 control unit 21 communication unit 22 storage unit 23 processing unit 24 SOC estimation unit 25 SOC representative value acquisition unit 26 learning data generation unit 27 learning model 28 learning processing unit 29 input data generation unit 50 deterioration simulator
Landscapes
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Chemical & Material Sciences (AREA)
- Electrochemistry (AREA)
- General Chemical & Material Sciences (AREA)
- Chemical Kinetics & Catalysis (AREA)
- Manufacturing & Machinery (AREA)
- Power Engineering (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Materials Engineering (AREA)
- Tests Of Electric Status Of Batteries (AREA)
- Secondary Cells (AREA)
- Charge And Discharge Circuits For Batteries Or The Like (AREA)
Abstract
Description
以下、本実施の形態に係る劣化推定装置を図面に基づいて説明する。図1は本実施の形態の遠隔監視システム100の概要を示す図である。図1に示すように、公衆通信網(例えば、インターネットなど)N1及び移動通信規格による無線通信を実現するキャリアネットワークN2などを含むネットワークNには、火力発電システムF、メガソーラー発電システムS、風力発電システムW、無停電電源装置(UPS:Uninterruptible Power Supply)U及び鉄道用の安定化電源システム等に配設される整流器(直流電源装置、又は交流電源装置)Dなどが接続されている。また、ネットワークNには、後述の通信デバイス1、通信デバイス1から情報を収集し、劣化推定装置としてのサーバ装置2、及び収集された情報を取得するクライアント装置3などが接続されている。本実施の形態において、劣化推定装置は、寿命シミュレータであってもよい。
作する。SOHは、現在の満充電容量を、新品で所定温度のときの満充電容量を基準に比較した指標である。例えば、SOHが80%は、新品の80%しか容量がないことを意味する。また、容量の代わりに、充放電可能な電力量を基準としてもよい。一般的には、SOHが閾値より小さくなると寿命に達し、蓄電素子は使用できないと判断される。
上述の第1実施形態では、SOCの代表値という抽象化されたデータを用いる構成であったが、これに限定されるものではない。第2実施形態では、抽象化の度合いを小さくして、よりセンサ情報などの時系列データに近い態様で学習用データを生成することができる。
上述の第1実施形態及び第2実施形態では、学習処理部28(処理部23)が、学習モデル27を学習及び更新する構成であったが、第3実施形態では、学習処理部28(処理部23)は、後述の劣化シミュレータが出力する蓄電素子の劣化値(SOH)を学習データとして用いて学習モデル27を学習させ、第1実施形態及び第2実施形態と同様の処理を用いて学習済の学習モデル27を再学習させることができる。以下、第3実施形態について説明する。
20 制御部
21 通信部
22 記憶部
23 処理部
24 SOC推定部
25 SOC代表値取得部
26 学習データ生成部
27 学習モデル
28 学習処理部
29 入力データ生成部
50 劣化シミュレータ
Claims (21)
- 蓄電素子の劣化を推定する劣化推定装置であって、
蓄電素子の第1時点でのSOH及び前記第1時点より後の第2時点でのSOHを取得するSOH取得部と、
前記第1時点から第2時点までの間の前記蓄電素子のSOCの代表値を取得する代表値取得部と、
前記第1時点でのSOH及び前記代表値を入力データとし、前記第2時点でのSOHを出力データとする学習データに基づいて学習モデルを学習させる学習処理部と
を備える劣化推定装置。 - 前記第1時点から第2時点までの間の前記蓄電素子のSOCの推移を推定するSOC推定部を備え、
前記代表値取得部は、
前記SOCの推移に基づいて前記第1時点から第2時点までの間の、SOCの平均、SOCの総変動量、及びSOCの変動幅の少なくともいずれか一つを取得する請求項1に記載の劣化推定装置。 - 前記第1時点から第2時点までの間の前記蓄電素子のSOCの推移を推定するSOC推定部を備え、
前記代表値取得部は、
前記SOCの推移を前記代表値として取得する請求項1に記載の劣化推定装置。 - 前記第1時点から第2時点までの間の前記蓄電素子の温度の推移を取得する温度取得部を備え、
前記学習処理部は、
前記温度の推移をさらなる入力データとする学習データに基づいて学習モデルを学習させる請求項1から請求項3のいずれか一項に記載の劣化推定装置。 - 前記温度取得部は、
前記第1時点から第2時点までの間の前記蓄電素子の温度の代表値を取得し、
前記学習処理部は、
前記温度の代表値をさらなる入力データとする学習データに基づいて学習モデルを学習させる請求項1から請求項4のいずれか一項に記載の劣化推定装置。 - 前記学習処理部は、
前記蓄電素子の製造時点から前記第1時点までの経過期間をさらなる入力データとする学習データに基づいて学習モデルを学習させる請求項1から請求項5のいずれか一項に記載の劣化推定装置。 - 前記学習処理部は、
前記蓄電素子の製造時点から前記第1時点までの充放電回数をさらなる入力データとする学習データに基づいて学習モデルを学習させる請求項1から請求項6のいずれか一項に記載の劣化推定装置。 - 前記学習処理部は、
前記第1時点から第2時点までの学習期間を前記蓄電素子の使用期間に亘って複数設けて学習データに基づいて学習モデルを学習させる請求項1から請求項7のいずれか一項に記載の劣化推定装置。 - 前記学習処理部が学習させた学習済の学習モデルを用いて前記蓄電素子の劣化を推定する請求項1から請求項8のいずれか一項に記載の劣化推定装置。
- 蓄電素子の劣化を推定する劣化推定装置であって、
蓄電素子の第1時点でのSOHを取得するSOH取得部と、
前記第1時点から第2時点までの間の前記蓄電素子のSOCの代表値を取得する代表値取得部と、
前記第1時点でのSOH及び前記代表値を入力データとし、前記第2時点でのSOHを推定する学習済の学習モデルと
を備える劣化推定装置。 - 蓄電素子の劣化を推定する劣化推定装置であって、
蓄電素子のSOCの変動に基づいて前記蓄電素子のSOHを推定する劣化シミュレータに、SOCの代表値を入力し、前記劣化シミュレータが出力するSOHを取得する出力値取得部と、
前記劣化シミュレータに入力したSOCの代表値を取得する入力値取得部と、
前記入力値取得部で取得したSOCの代表値及び前記出力値取得部で取得したSOHを学習データとして用いて学習モデルを学習させる学習処理部と、
蓄電素子のSOCの代表値を取得する代表値取得部と、
前記蓄電素子のSOHを取得するSOH取得部と、
前記代表値取得部で取得したSOCの代表値及び前記SOH取得部で取得したSOHを学習データとして用いて前記学習処理部で学習させた前記学習モデルを再学習させる再学習処理部と
を備える劣化推定装置。 - 前記SOH取得部は、
蓄電素子の第1時点でのSOH及び前記第1時点より後の第2時点でのSOHを取得し、
前記代表値取得部は、
前記第1時点から第2時点までの間の前記蓄電素子のSOCの代表値を取得する請求項11に記載の劣化推定装置。 - 蓄電素子の劣化を推定する劣化推定装置であって、
蓄電素子のSOCの変動に基づいて前記蓄電素子のSOHを推定する劣化シミュレータに入力するSOCの代表値、及び前記SOCの代表値を前記劣化シミュレータに入力したときに前記劣化シミュレータが出力するSOHを学習データとして学習させた学習モデルを備え、
前記学習モデルは、さらに、蓄電素子のSOCの代表値及び前記蓄電素子のSOHを学習データとして再学習してあり、
蓄電素子の第1時点でのSOHを取得するSOH取得部と、
前記第1時点から第2時点までの間の前記蓄電素子のSOCの代表値を取得する代表値取得部と
を備え、
前記第1時点でのSOH及び前記代表値取得部で取得した代表値を前記学習モデルに入力して前記第2時点でのSOHを推定する劣化推定装置。 - コンピュータに、蓄電素子の劣化を推定させるためのコンピュータプログラムであって、
コンピュータに、
蓄電素子の第1時点でのSOH及び前記第1時点より後の第2時点でのSOHを取得する処理と、
前記第1時点から第2時点までの間の前記蓄電素子のSOCの代表値を取得する処理と、
前記第1時点でのSOH及び前記代表値を入力データとし、前記第2時点でのSOHを出力データとする学習データに基づいて学習モデルを学習させる処理と
を実行させるコンピュータプログラム。 - コンピュータに、蓄電素子の劣化を推定させるためのコンピュータプログラムであって、
コンピュータに、
蓄電素子の第1時点でのSOHを取得する処理と、
前記第1時点から第2時点までの間の前記蓄電素子のSOCの代表値を取得する処理と、
前記第1時点でのSOH及び前記代表値を学習済の学習モデルに入力して前記第2時点でのSOHを推定する処理と
を実行させるコンピュータプログラム。 - コンピュータに、蓄電素子の劣化を推定させるためのコンピュータプログラムであって、
コンピュータに、
蓄電素子のSOCの変動に基づいて前記蓄電素子のSOHを推定する劣化シミュレータに、SOCの代表値が入力されて、前記劣化シミュレータが出力するSOHを取得する処理と、
前記劣化シミュレータに入力したSOCの代表値を取得する処理と、
取得したSOCの代表値及び取得したSOHを学習データとして用いて学習モデルを学習させる処理と、
蓄電素子のSOCの代表値を取得する処理と、
前記蓄電素子のSOHを取得する処理と、
取得したSOCの代表値及び取得したSOHを学習データとして用いて、学習させた前記学習モデルを再学習させる処理と
を実行させるコンピュータプログラム。 - コンピュータに、蓄電素子の劣化を推定させるためのコンピュータプログラムであって、
コンピュータに、
蓄電素子の第1時点でのSOHを取得する処理と、
前記第1時点から第2時点までの間の前記蓄電素子のSOCの代表値を取得する処理と、
前記第1時点でのSOH及び前記代表値を、蓄電素子のSOCの変動に基づいて前記蓄電素子のSOHを推定する劣化シミュレータに入力するSOCの代表値、及び前記SOCの代表値を前記劣化シミュレータに入力したときに前記劣化シミュレータが出力するSOHを学習データとして学習させ、さらに、蓄電素子のSOCの代表値及び前記蓄電素子のSOHを学習データとして再学習させた学習モデルに入力して前記第2時点でのSOHを推定する処理と
を実行させるコンピュータプログラム。 - 蓄電素子の劣化を推定する劣化推定方法であって、
蓄電素子の第1時点でのSOH及び前記第1時点より後の第2時点でのSOHを取得し、
前記第1時点から第2時点までの間の前記蓄電素子のSOCの代表値を取得し、
前記第1時点でのSOH及び前記代表値を入力データとし、前記第2時点でのSOHを出力データとする学習データに基づいて学習モデルを学習させる劣化推定方法。 - 蓄電素子の劣化を推定する劣化推定方法であって、
蓄電素子の第1時点でのSOHを取得し、
前記第1時点から第2時点までの間の前記蓄電素子のSOCの代表値を取得し、
前記第1時点でのSOH及び前記代表値を学習済の学習モデルに入力して前記第2時点でのSOHを推定する劣化推定方法。 - 蓄電素子の劣化を推定する劣化推定方法であって、
蓄電素子のSOCの変動に基づいて前記蓄電素子のSOHを推定する劣化シミュレータに、SOCの代表値を入力し、前記劣化シミュレータが出力するSOHを取得し、
前記劣化シミュレータに入力したSOCの代表値を取得し、
取得されたSOCの代表値及び取得されたSOHを学習データとして用いて学習モデルを学習させ、
蓄電素子のSOCの代表値を取得し、
前記蓄電素子のSOHを取得し、
取得されたSOCの代表値及び取得されたSOHを学習データとして用いて、学習させた前記学習モデルを再学習させる劣化推定方法。 - 蓄電素子の劣化を推定する劣化推定方法であって、
蓄電素子の第1時点でのSOHを取得し、
前記第1時点から第2時点までの間の前記蓄電素子のSOCの代表値を取得し、
前記第1時点でのSOH及び前記代表値を、蓄電素子のSOCの変動に基づいて前記蓄電素子のSOHを推定する劣化シミュレータに入力するSOCの代表値、及び前記SOCの代表値を前記劣化シミュレータに入力したときに前記劣化シミュレータが出力するSOHを学習データとして学習させ、さらに、蓄電素子のSOCの代表値及び前記蓄電素子のSOHを学習データとして再学習させた学習モデルに入力して前記第2時点でのSOHを推定する劣化推定方法。
Priority Applications (4)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| AU2019238652A AU2019238652A1 (en) | 2018-03-20 | 2019-03-14 | Degradation estimation apparatus, computer program, and degradation estimation method |
| EP19770833.2A EP3770619B1 (en) | 2018-03-20 | 2019-03-14 | Degradation estimating device and degradation estimating method |
| US16/981,855 US20210033675A1 (en) | 2018-03-20 | 2019-03-14 | Degradation estimation apparatus, computer program, and degradation estimation method |
| CN201980018299.4A CN111868539A (zh) | 2018-03-20 | 2019-03-14 | 劣化估计装置、计算机程序以及劣化估计方法 |
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2018-053016 | 2018-03-20 | ||
| JP2018053016 | 2018-03-20 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2019181728A1 true WO2019181728A1 (ja) | 2019-09-26 |
Family
ID=67987641
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/JP2019/010543 Ceased WO2019181728A1 (ja) | 2018-03-20 | 2019-03-14 | 劣化推定装置、コンピュータプログラム及び劣化推定方法 |
Country Status (6)
| Country | Link |
|---|---|
| US (1) | US20210033675A1 (ja) |
| EP (1) | EP3770619B1 (ja) |
| JP (1) | JP6579287B1 (ja) |
| CN (1) | CN111868539A (ja) |
| AU (1) | AU2019238652A1 (ja) |
| WO (1) | WO2019181728A1 (ja) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3812783A1 (en) * | 2019-10-23 | 2021-04-28 | Novum engineerING GmbH | Estimating a battery state from electrical impedance measurements using convolutional neural network means |
| WO2021210526A1 (ja) | 2020-04-17 | 2021-10-21 | 株式会社エンビジョンAescジャパン | 残容量推定装置、モデル生成装置、残容量推定方法、モデル生成方法、及びプログラム |
| WO2021241115A1 (ja) | 2020-05-25 | 2021-12-02 | 株式会社エンビジョンAescジャパン | 劣化推定装置、モデル生成装置、劣化推定方法、モデル生成方法、及びプログラム |
| JP2024029489A (ja) * | 2022-08-22 | 2024-03-06 | 株式会社デンソー | 充放電制御システム及び充放電制御プログラム |
| JP2024083789A (ja) * | 2022-12-12 | 2024-06-24 | いすゞ自動車株式会社 | 劣化推定装置 |
Families Citing this family (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR102781907B1 (ko) * | 2019-11-28 | 2025-03-18 | 주식회사 엘지에너지솔루션 | 배터리 셀 이상 퇴화 진단 장치 및 방법 |
| JP7413806B2 (ja) * | 2020-02-06 | 2024-01-16 | トヨタ自動車株式会社 | バッテリ劣化判定装置、バッテリ劣化判定方法、及びバッテリ劣化判定プログラム |
| DE102020212298A1 (de) * | 2020-09-29 | 2022-03-31 | Robert Bosch Gesellschaft mit beschränkter Haftung | Verfahren und Vorrichtung zur gerätespezifischen Bestimmung eines innenwiderstandsbasierten Alterungszustands einer Batterie |
| KR102829175B1 (ko) * | 2020-10-08 | 2025-07-07 | 주식회사 엘지에너지솔루션 | 배터리 퇴화도 산출 방법 및 배터리 퇴화도 산출 장치 |
| DE102020215204A1 (de) | 2020-12-02 | 2022-06-02 | Robert Bosch Gesellschaft mit beschränkter Haftung | Verfahren und Vorrichtung zur Vorhersage eines Alterungszustands einer Gerätebatterie in einem batteriebetriebenen Gerät |
| JP7344192B2 (ja) * | 2020-12-21 | 2023-09-13 | 矢崎総業株式会社 | バッテリ制御装置 |
| JP7730645B2 (ja) * | 2021-02-04 | 2025-08-28 | 株式会社野村総合研究所 | 情報処理装置、情報処理方法、プログラム及びデータ構造 |
| JP7604992B2 (ja) * | 2021-03-29 | 2024-12-24 | 株式会社Gsユアサ | 推定支援装置、コンピュータプログラム、推定支援方法及び学習済みニューラルネットワークシステムの生成方法 |
| JP2022157734A (ja) * | 2021-03-31 | 2022-10-14 | 本田技研工業株式会社 | 劣化状態推定システム、劣化状態推定方法、及びプログラム |
| JP2022157740A (ja) * | 2021-03-31 | 2022-10-14 | 本田技研工業株式会社 | 状態推定システム、状態推定方法、及びプログラム |
| KR20220168920A (ko) * | 2021-06-17 | 2022-12-26 | 주식회사 엘지에너지솔루션 | 배터리 soh 추정 장치 및 방법 |
| US20220416548A1 (en) * | 2021-06-23 | 2022-12-29 | TWAICE Technologies GmbH | Operational planning for battery-based energy storage systems considering battery aging |
| US12206283B2 (en) * | 2021-08-20 | 2025-01-21 | Vertiv Corporation | Systems and methods for operating an uninterruptible power supply (UPS) |
| JP7366975B2 (ja) | 2021-09-30 | 2023-10-23 | 本田技研工業株式会社 | バッテリ劣化推定装置、バッテリ劣化推定システム、バッテリ劣化推定方法、およびプログラム |
| US20240410956A1 (en) * | 2021-10-14 | 2024-12-12 | Gs Yuasa International Ltd. | Lead-acid storage device, information processing method, and computer program |
| KR20230140743A (ko) * | 2022-03-30 | 2023-10-10 | 주식회사 엘지에너지솔루션 | 필드 데이터 기반 인공신경망 배터리 모델을 통한 배터리 이상 검출 방법 및 시스템 |
| JP2024025468A (ja) * | 2022-08-12 | 2024-02-26 | 株式会社東芝 | 状態推定方法、状態推定装置、状態推定システム及び状態推定プログラム |
| KR102888922B1 (ko) * | 2022-08-30 | 2025-11-20 | 주식회사 엘지에너지솔루션 | 배터리 상태 예측 장치 및 그것의 동작 방법 |
| US12493078B2 (en) | 2023-10-11 | 2025-12-09 | Zitara Technologies, Inc. | System and method for degradation based battery control |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH09243716A (ja) * | 1996-03-08 | 1997-09-19 | Sanken Electric Co Ltd | 二次電池の状態判定方法 |
| JP2001325582A (ja) * | 2000-05-17 | 2001-11-22 | Chugoku Electric Power Co Inc:The | 時系列データ学習・予測装置 |
| JP2006220616A (ja) * | 2005-02-14 | 2006-08-24 | Denso Corp | 車両用蓄電装置の内部状態検出方式 |
| JP2008232758A (ja) | 2007-03-19 | 2008-10-02 | Nippon Soken Inc | 二次電池の内部状態検出装置及びニューラルネット式状態量推定装置 |
| WO2016092811A1 (ja) * | 2014-12-10 | 2016-06-16 | 株式会社Gsユアサ | 蓄電素子状態推定装置及び蓄電素子状態推定方法 |
| JP2016223964A (ja) * | 2015-06-02 | 2016-12-28 | 三菱電機株式会社 | 蓄電デバイスの劣化診断装置及び劣化診断方法 |
| WO2017002292A1 (ja) * | 2015-06-30 | 2017-01-05 | ソニー株式会社 | 蓄電システム、コントローラおよび蓄電池の充放電方法 |
| JP2017065532A (ja) | 2015-09-30 | 2017-04-06 | ヤマハ発動機株式会社 | 車両 |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE602006002896D1 (de) * | 2005-02-14 | 2008-11-13 | Denso Corp | Verfahren und Vorrichtung zur Erkennung des Ladestatus einer Sekundärbatterie auf Basis neuronaler Netzwerkkalkulation |
| FR2943794B1 (fr) * | 2009-03-24 | 2011-05-06 | Saft Groupe Sa | Procede de determination de l'etat de sante d'une batterie |
| WO2016208251A1 (ja) * | 2015-06-25 | 2016-12-29 | 株式会社日立製作所 | 蓄電システム |
| KR102553031B1 (ko) * | 2016-01-14 | 2023-07-06 | 삼성전자주식회사 | 배터리의 상태 추정 장치 및 방법 |
| KR101846913B1 (ko) * | 2016-11-01 | 2018-04-09 | 현대자동차 주식회사 | 하이브리드 차량의 배터리 충전 제어 장치 및 충전 제어 방법 |
-
2019
- 2019-03-14 US US16/981,855 patent/US20210033675A1/en not_active Abandoned
- 2019-03-14 CN CN201980018299.4A patent/CN111868539A/zh active Pending
- 2019-03-14 EP EP19770833.2A patent/EP3770619B1/en active Active
- 2019-03-14 AU AU2019238652A patent/AU2019238652A1/en not_active Abandoned
- 2019-03-14 WO PCT/JP2019/010543 patent/WO2019181728A1/ja not_active Ceased
- 2019-03-18 JP JP2019050400A patent/JP6579287B1/ja active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH09243716A (ja) * | 1996-03-08 | 1997-09-19 | Sanken Electric Co Ltd | 二次電池の状態判定方法 |
| JP2001325582A (ja) * | 2000-05-17 | 2001-11-22 | Chugoku Electric Power Co Inc:The | 時系列データ学習・予測装置 |
| JP2006220616A (ja) * | 2005-02-14 | 2006-08-24 | Denso Corp | 車両用蓄電装置の内部状態検出方式 |
| JP2008232758A (ja) | 2007-03-19 | 2008-10-02 | Nippon Soken Inc | 二次電池の内部状態検出装置及びニューラルネット式状態量推定装置 |
| WO2016092811A1 (ja) * | 2014-12-10 | 2016-06-16 | 株式会社Gsユアサ | 蓄電素子状態推定装置及び蓄電素子状態推定方法 |
| JP2016223964A (ja) * | 2015-06-02 | 2016-12-28 | 三菱電機株式会社 | 蓄電デバイスの劣化診断装置及び劣化診断方法 |
| WO2017002292A1 (ja) * | 2015-06-30 | 2017-01-05 | ソニー株式会社 | 蓄電システム、コントローラおよび蓄電池の充放電方法 |
| JP2017065532A (ja) | 2015-09-30 | 2017-04-06 | ヤマハ発動機株式会社 | 車両 |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3812783A1 (en) * | 2019-10-23 | 2021-04-28 | Novum engineerING GmbH | Estimating a battery state from electrical impedance measurements using convolutional neural network means |
| WO2021210526A1 (ja) | 2020-04-17 | 2021-10-21 | 株式会社エンビジョンAescジャパン | 残容量推定装置、モデル生成装置、残容量推定方法、モデル生成方法、及びプログラム |
| US12443152B2 (en) | 2020-04-17 | 2025-10-14 | Aesc Japan Ltd. | Remaining capacity estimation apparatus, model generation apparatus, and non-transitory computer-readable medium |
| WO2021241115A1 (ja) | 2020-05-25 | 2021-12-02 | 株式会社エンビジョンAescジャパン | 劣化推定装置、モデル生成装置、劣化推定方法、モデル生成方法、及びプログラム |
| JP2021185354A (ja) * | 2020-05-25 | 2021-12-09 | 株式会社エンビジョンAescジャパン | 劣化推定装置、モデル生成装置、劣化推定方法、モデル生成方法、及びプログラム |
| JP7457575B2 (ja) | 2020-05-25 | 2024-03-28 | 株式会社Aescジャパン | 劣化推定装置、モデル生成装置、劣化推定方法、モデル生成方法、及びプログラム |
| JP2024029489A (ja) * | 2022-08-22 | 2024-03-06 | 株式会社デンソー | 充放電制御システム及び充放電制御プログラム |
| JP2024083789A (ja) * | 2022-12-12 | 2024-06-24 | いすゞ自動車株式会社 | 劣化推定装置 |
Also Published As
| Publication number | Publication date |
|---|---|
| US20210033675A1 (en) | 2021-02-04 |
| EP3770619A4 (en) | 2021-04-28 |
| EP3770619B1 (en) | 2023-10-25 |
| AU2019238652A1 (en) | 2020-11-12 |
| JP2019168452A (ja) | 2019-10-03 |
| CN111868539A (zh) | 2020-10-30 |
| JP6579287B1 (ja) | 2019-09-25 |
| EP3770619A1 (en) | 2021-01-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP6579287B1 (ja) | 劣化推定装置、コンピュータプログラム及び劣化推定方法 | |
| JP6620902B2 (ja) | 劣化推定装置、コンピュータプログラム及び劣化推定方法 | |
| AU2019238615B2 (en) | Abnormality factor determination apparatus, degradation determination apparatus, computer program, degradation determining method, and abnormality factor determining method | |
| JP6590029B1 (ja) | 行動生成装置、蓄電素子評価装置、コンピュータプログラム、学習方法及び評価方法 | |
| CN113589189B (zh) | 基于充放电数据特征的锂电池健康状况预测方法及装置 | |
| JP7298114B2 (ja) | 状態推定方法、及び状態推定装置 | |
| WO2019203111A1 (ja) | 状態推定方法、及び状態推定装置 | |
| US20220113354A1 (en) | Data processing apparatus, data processing method and computer readable medium | |
| Savargaonkar et al. | A novel neural network with Gaussian process feedback for modeling the state-of-charge of battery cells | |
| JP6973099B2 (ja) | 監視装置、監視方法及びコンピュータプログラム | |
| Sui et al. | Early prediction of lithium-ion batteries lifetime via few-shot learning | |
| Senthilkumar et al. | A Novel Hybrid Approach for Remaining Useful Life (RUL) and Short-Term Capacity Prediction of Batteries | |
| Khobre et al. | LSTM-Based SOC Estimation Leveraging Coulomb Counting: A Machine Learning Approach | |
| Manjunatha Babu et al. | A Hybrid Technique for Estimating the State of Health of Lithium‐Ion Batteries | |
| JP2025117337A (ja) | 電力貯蔵装置監視システムおよび電力貯蔵装置監視方法 |
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: 19770833 Country of ref document: EP Kind code of ref document: A1 |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| ENP | Entry into the national phase |
Ref document number: 2019770833 Country of ref document: EP Effective date: 20201020 |
|
| ENP | Entry into the national phase |
Ref document number: 2019238652 Country of ref document: AU Date of ref document: 20190314 Kind code of ref document: A |