Manufacturing method, manufacturing device, and program

Abstract

The manufacturing apparatus acquires a plurality of first log data items indicating the state of a battery during charging or discharging acquired from a battery mounting apparatus, and a plurality of first degradation degrees of the battery, and calculates the reliability of each of the first degradation degrees on the basis of the content of charging or discharging corresponding to each of the first log data items. And outputs a model for which the estimation accuracy of the degree of deterioration of the battery is optimal, from among a model in which machine learning is performed on the relationship between the plurality of first degrees of deterioration and the plurality of first log data, and a model in which machine learning is performed on the relationship between the first degree of deterioration having a degree of reliability equal to or greater than a predetermined value and the first log data corresponding to the first degree of deterioration.

Description

Technical Field

[0001] The present disclosure relates to a technology for generating a learned model for estimating the degree of degradation of a chargeable and dischargeable battery. Background Art

[0002] In the past, the following technology is known: the state of a battery that can be charged and discharged is measured in a test environment, and the degree of degradation of the battery is estimated using a learning model generated by machine learning of the measured battery state. For example, Patent Document 1 states: a prediction model is constructed by learning the relationship between the current, voltage, and temperature of a sample battery and the battery capacity for each charge and discharge cycle, and the degree of degradation of the sample battery is estimated based on the battery capacity predicted using the prediction model and the rated capacity.

[0003] However, in Patent Document 1, since the state of a sample battery needs to be measured at each charge and discharge cycle, there is still room for improvement in terms of quickly generating a prediction model with high accuracy.

[0004] Prior Art Literature

[0005] Patent Literature

[0006] Patent Document 1: Chinese Patent Application Publication No. 112051511 Summary of the invention

[0007] The present disclosure has been made to solve the above-mentioned problems, and an object of the present disclosure is to provide a technology capable of quickly generating a learned model capable of estimating the degree of battery degradation with high accuracy.

[0008] Means for solving problems

[0009] A manufacturing method according to one aspect of the present disclosure is a manufacturing method in a manufacturing device having a learning model for estimating the degree of degradation of a chargeable and dischargeable battery, wherein a plurality of first log data indicating the state of the battery during charging or discharging obtained from a device carrying the battery and a plurality of first degradation degrees indicating the degree of degradation of the battery calculated by a degradation degree estimation method using each of the first log data are obtained, and based on the contents of the charge or discharge corresponding to each of the first log data, i.e., the first charge and discharge, the reliability of the certainty of the first degradation degree corresponding to each of the first log data is calculated, and the reliability of the certainty of the first degradation degree corresponding to each of the first log data is calculated by estimating the first degradation degree of the battery during charging or discharging. A first learned model is generated by performing machine learning on the relationship between the first degradation degree and the multiple first log data, a second learned model is generated by performing machine learning on the relationship between the first degradation degree whose reliability is above a given value among the multiple first degradation degrees and the first log data corresponding to the first degradation degree, the estimated accuracy of the degradation degree of the battery in each of the first learned model and the second learned model is evaluated, and the learned model evaluated as having the best estimation accuracy among the first learned model and the second learned model is output. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 It is a diagram showing the overall configuration of a model manufacturing system in the first embodiment of the present disclosure.

[0011] Figure 2 This is a diagram showing an example of the relationship between the content of charge and discharge and the coefficient.

[0012] Figure 3 This is a diagram showing an example of feature quantities used as explanatory variables for machine learning.

[0013] Figure 4 This is a flowchart showing an example of the model making process.

[0014] Figure 5 This is a diagram showing an example of a process of extracting the first degree of degradation within a given deviation range. DETAILED DESCRIPTION

[0015] (Insights that form the basis of this disclosure)

[0016] In the above-mentioned prior art, there are problems such as the cost of time in units of years required to obtain data used in machine learning in a test environment and the operating cost of the test environment. For this reason, in recent years, a degradation degree estimation method has been proposed, which obtains log data representing the state of the battery when it is actually charged or discharged from a device equipped with a battery such as an electric car, and uses the log data to calculate the degradation degree of the battery. However, although the degradation degree estimation method can quickly obtain the data used in calculating the degradation degree of the battery, there is a problem that the calculation accuracy of the degradation degree of the battery is low.

[0017] For example, in the two-point OCV estimation method widely known as a method for estimating the degree of degradation, data indicating the voltage of the battery when the battery is in a rest state is obtained immediately before and immediately after charging and discharging, as data indicating the open circuit voltage of the battery, and the degree of degradation of the battery is calculated using the open circuit voltage. However, due to reasons such as the short period of time during which the battery is in a rest state, data indicating the voltage of the battery when the chemical reaction inside the battery has not yet ended is sometimes obtained as data indicating the open circuit voltage of the battery. In this case, the degree of degradation of the battery may not be accurately calculated using the open circuit voltage.

[0018] To this end, the present inventors conducted intensive research on a technique capable of rapidly generating a learned model capable of estimating the degree of battery degradation with high accuracy, and have arrived at the respective aspects corresponding to the present disclosure shown below.

[0019] (1) A manufacturing method according to one aspect of the present disclosure is a manufacturing method in a manufacturing device having a learned model for estimating the degree of degradation of a chargeable and dischargeable battery, wherein a plurality of first log data indicating the state of the battery during charging or discharging obtained from a device carrying the battery and a plurality of first degradation degrees indicating the degree of degradation of the battery calculated by a degradation degree estimation method using each of the first log data are obtained, and based on the contents of the charge or discharge corresponding to each of the first log data, i.e., the first charge and discharge, a reliability of certainty indicating the first degradation degree corresponding to each of the first log data is calculated, and the reliability of certainty indicating the first degradation degree corresponding to each of the first log data is calculated by estimating the first degradation degree of the battery during charging or discharging. A first learned model is generated by performing machine learning on the relationship between multiple first degradation degrees and the multiple first log data, a second learned model is generated by performing machine learning on the relationship between the first degradation degree whose reliability is above a given value among the multiple first degradation degrees and the first log data corresponding to the first degradation degree, the estimated accuracy of the degradation degree of the battery in each of the first learned model and the second learned model is evaluated, and the learned model evaluated as having the best estimation accuracy among the first learned model and the second learned model is output.

[0020] In this structure, a plurality of first log data indicating the state of the battery during charging or discharging is obtained from a device equipped with the battery. For this reason, this structure can more quickly obtain a plurality of first log data required for machine learning for generating the first learned model and the second learned model than when multiple charging and discharging tests are performed in a test environment. Thus, this structure can quickly generate the first learned model and the second learned model.

[0021] In addition, in the present configuration, a second learned model is generated by performing machine learning on the relationship between the first degree of degradation having a reliability greater than a given value and the first log data corresponding to the first degree of degradation. For this reason, the present configuration can generate a second learned model that is considered to be able to estimate the degree of degradation of the battery more accurately than the first learned model that has the possibility of performing machine learning on the relationship between the first degree of degradation having a reliability less than a given value and the first log data corresponding to the first degree of degradation.

[0022] However, even the second learned model generated in this way does not necessarily have a higher estimation accuracy of the battery degradation degree than the first learned model. Therefore, in this structure, the estimation accuracy of the battery degradation degree in each of the first learned model and the second learned model is evaluated, and the learned model evaluated as having the best estimation accuracy is output. Therefore, this structure can output a learned model that can estimate the battery degradation degree with good accuracy.

[0023] (2) In the manufacturing method described in (1) above, it is also possible to further extract the first degradation degree that is within a given deviation range from among the multiple first degradation degrees, and further generate a third learned model by performing machine learning on the relationship between the extracted first degradation degree and the first log data corresponding to the first degradation degree. In the evaluation, the estimation accuracy in the third learned model is further evaluated, and in the output, the learned model evaluated as having the best estimation accuracy among the first learned model, the second learned model, and the third learned model is output.

[0024] In this structure, a third learned model is generated by performing machine learning on the relationship between the first degree of degradation within a given deviation range and the first log data corresponding to the first degree of degradation. Therefore, this structure can generate a third learned model that estimates the degree of degradation of the battery more accurately than the first learned model that is considered to have a possibility of performing machine learning on the relationship between the first degree of degradation outside the deviation range and the first log data corresponding to the first degree of degradation.

[0025] However, even the third learned model generated in this way does not necessarily have a higher estimation accuracy of the battery degradation degree than the first learned model and the second learned model. Therefore, in this structure, the estimation accuracy of the battery degradation degree of the third learned model is further evaluated, and the learned model evaluated as having the best estimation accuracy among the first learned model, the second learned model, and the third learned model is output. Therefore, this structure can output a learned model that can estimate the battery degradation degree with good accuracy.

[0026] (3) In the manufacturing method described in (2) above, a fourth learned model may be generated by further performing machine learning on the relationship between the first degradation degree whose reliability is greater than the given value among the extracted first degradation degrees and the first log data corresponding to the first degradation degree, wherein in the evaluation, the estimated accuracy in the fourth learned model is further evaluated, and in the output, the learned model evaluated as having the best estimated accuracy among the first learned model, the second learned model, the third learned model and the fourth learned model is output.

[0027] In this structure, a fourth learned model is generated by performing machine learning on the relationship between the first degradation degree having a reliability greater than a given value among the first degradation degrees within a given deviation range and the first log data corresponding to the first degradation degree. Therefore, this structure can generate a fourth learned model that estimates the battery degradation degree more accurately than the third learned model that is considered to have a possibility of performing machine learning on the relationship between the first degradation degree having a reliability less than a given value among the first degradation degrees within the deviation range and the first log data corresponding to the first degradation degree.

[0028] However, even the fourth learned model generated in this way does not necessarily have a higher estimation accuracy of the battery degradation degree than the first learned model, the second learned model, and the third learned model. Therefore, in this structure, the estimation accuracy of the battery degradation degree of the fourth learned model is further evaluated, and the learned model evaluated as having the best estimation accuracy among the first learned model, the second learned model, the third learned model, and the fourth learned model is output. Therefore, this structure can output a learned model that can estimate the battery degradation degree with good accuracy.

[0029] (4) In the manufacturing method described in any one of (1) to (3) above, it is also possible that, in the evaluation, second log data indicating the state of the battery during the second charging when the SOC of the battery is charged from 0% to 100% when the temperature of the battery is greater than 20 degrees and less than 30 degrees, and a second degradation degree indicating the degradation degree of the battery calculated by the degradation degree estimation method using the second log data are obtained from a device carrying the battery, and the degree of deviation between the degradation degree of the battery estimated by inputting the second log data into each learned model and the second degradation degree is calculated for each of the evaluation objects, i.e., multiple learned models, and the learned model with the smallest degree of deviation among the multiple learned models is evaluated as the learned model with the best estimation accuracy.

[0030] According to this configuration, the estimation accuracy of the degree of battery degradation in each of the plurality of learned models can be appropriately and quickly evaluated using the second log data indicating the state of the battery during the second charge acquired from the device equipped with the battery.

[0031] (5) In the manufacturing method described in (1) above, in calculating the reliability, a result obtained by multiplying an initial value of the reliability greater than the given value by a coefficient corresponding to the content of the first charge and discharge may be calculated as the reliability.

[0032] According to this configuration, the reliability of the first degree of degradation corresponding to each first log data can be appropriately calculated using a coefficient according to the content of the first charge and discharge.

[0033] (6) In the manufacturing method described in (5) above, the coefficient may be determined as a difference between the SOC of the battery at the start of the first charge / discharge and the SOC of the battery at the end of the first charge / discharge.

[0034] In this structure, the smaller the difference between the SOC of the battery at the beginning of the first charge and discharge and the SOC of the battery at the end of the first charge and discharge, the smaller the reliability calculated. Therefore, in this structure, the smaller the difference between the SOC of the battery at the beginning of the first charge and discharge and the SOC of the battery at the end, the less likely it is that the first log data corresponding to the first charge and discharge will be machine-learned in order to generate the second learning model.

[0035] (7) In the manufacturing method described in (5) or (6) above, when calculating the reliability, if the time during which the battery is in a rest state immediately before the first charge / discharge is less than a predetermined time, a coefficient less than 1 may be multiplied.

[0036] In this structure, when the time when the battery is in a rest state immediately before the first charge and discharge is less than a given time, a reliability lower than an initial value is calculated. Therefore, this structure can reduce the possibility of machine learning for the first log data corresponding to the first charge and discharge immediately after the battery is in a rest state for less than a given time in order to generate a second learned model.

[0037] (8) In the manufacturing method described in any one of (5) to (7) above, in calculating the reliability, when the first degree of degradation corresponding to each first log data is greater than a given upper limit value or less than a given lower limit value, it may be multiplied by a coefficient less than 1.

[0038] In this structure, when the first degradation degree corresponding to each first log data is greater than a given upper limit value, or less than a given lower limit value, a reliability lower than an initial value is calculated. Therefore, when the first degradation degree greater than the upper limit value or less than the lower limit value is calculated using the first log data, the possibility of machine learning on the first degradation degree and the first log data in order to generate the second learned model can be reduced.

[0039] (9) In the manufacturing method described in (1) above, in the method for estimating the degree of degradation, the voltage of the battery when the battery is in a rest state is obtained immediately before and immediately after the first charge and discharge, as the open circuit voltage of the battery immediately before and immediately after the first charge and discharge, and the open circuit voltage of the battery immediately before and immediately after the first charge and discharge is determined as the SOC of the battery at the start and end of the first charge and discharge with reference to information indicating the relationship between the SOC of the battery and the open circuit voltage of the battery, and the difference between the SOC of the battery at the start and end of the first charge and discharge is calculated, and the integrated value of the current of the battery during the first charge and discharge is calculated using each first log data, and the result obtained by dividing the integrated value by the difference by the full charge capacity of the battery in the initial state is calculated as the degree of degradation of the battery.

[0040] According to this configuration, the first degree of degradation can be quickly calculated using each first log data acquired from the device equipped with the battery and the full charge capacity of the battery in the initial state.

[0041] (10) In the manufacturing method described in (4) above, in the method for estimating the degree of degradation, the voltage of the battery when the battery is in a rest state is obtained immediately before and immediately after the second charge is performed, as the open circuit voltage of the battery immediately before and immediately after the second charge is performed, and the open circuit voltage of the battery immediately before and immediately after the second charge is performed is determined as the SOC of the battery at the start and end of the second charge with reference to information indicating the relationship between the SOC of the battery and the open circuit voltage of the battery, and the difference between the SOC of the battery at the start and end of the second charge is calculated, and the integrated value of the current of the battery during the second charge is calculated using the second log data, and the result obtained by dividing the integrated value by the difference by the full charge capacity of the battery in the initial state is calculated as the degree of degradation of the battery.

[0042] According to this configuration, the second degree of degradation can be quickly calculated using each second log data acquired from the device equipped with the battery and the full charge capacity of the battery in the initial state.

[0043] (11) In the manufacturing method described in (2) or (3) above, in extracting the first degree of degradation within the deviation range, a linear regression is performed using the square root of the time from the time when the battery is in the initial state to the time when the first charge and discharge is completed as an explanatory variable and each first degree of degradation as a target variable, and the first degree of degradation whose distance to the regression line obtained by the linear regression is less than a given distance from the multiple first degrees of degradation is extracted as the first degree of degradation within the deviation range.

[0044] According to this configuration, the first degradation degree within a given deviation range can be appropriately extracted from among a plurality of first degradation degrees using the regression line obtained by performing the linear regression.

[0045] (12) A manufacturing device according to another aspect of the present disclosure is a manufacturing device having a learned model for estimating the degree of degradation of a chargeable and dischargeable battery, and comprises: an acquisition unit for extracting a plurality of first log data indicating the state of the battery during charging or discharging, obtained from a device carrying the battery, and a plurality of first degradation degrees indicating the degree of degradation of the battery calculated by a degradation degree estimation method using each of the first log data; a calculation unit for calculating the reliability of the certainty of the first degradation degree corresponding to each of the first log data based on the content of the charge or discharge corresponding to each of the first log data, i.e., the first charge and discharge; and a first generation unit for generating a first degradation degree estimation model based on the first log data. A first learning model is generated by performing machine learning on the relationship between a first degradation degree and the plurality of first log data; a second generating unit is generated by performing machine learning on the relationship between a first degradation degree having a reliability greater than a given value among the plurality of first degradation degrees and the first log data corresponding to the first degradation degree to generate a second learning model; an evaluating unit is used to evaluate the estimated accuracy of the battery degradation degree in each of the first learning model and the second learning model; and an output unit is used to output the learning model evaluated as having the best estimation accuracy among the first learning model and the second learning model.

[0046] According to this structure, the same effects as those of the manufacturing method described in the above (1) can be obtained.

[0047] (13) A program according to another aspect of the present disclosure is a program for a manufacturing device that estimates a learned model of the degree of degradation of a chargeable and dischargeable battery, causing the manufacturing device to perform the following processing: obtaining a plurality of first log data indicating the state of the battery during charging or discharging obtained from a device on which the battery is mounted and a plurality of first degradation degrees indicating the degree of degradation of the battery calculated by a degradation degree estimation method using each first log data, and calculating a reliability indicating the certainty of the first degradation degree corresponding to each first log data based on the content of the charge or discharge corresponding to each first log data, i.e., the first charge and discharge; A first learned model is generated by performing machine learning on the relationship between the multiple first degradation degrees and the multiple first log data, and a second learned model is generated by performing machine learning on the relationship between the first degradation degree whose reliability is above a given value among the multiple first degradation degrees and the first log data corresponding to the first degradation degree, and the estimated accuracy of the degradation degree of the battery in each of the first learned model and the second learned model is evaluated, and the learned model evaluated as having the best estimation accuracy among the first learned model and the second learned model is output.

[0048] According to this structure, the same effects as those of the manufacturing method described in the above (1) can be obtained.

[0049] The present disclosure can also be realized as a system that operates according to such a program. In addition, it is needless to say that such a computer program can be distributed via a computer-readable non-transitory recording medium such as a CD-ROM or a communication network such as the Internet.

[0050] In addition, the embodiments described below all represent a specific example of the present disclosure. The numerical values, shapes, constituent elements, steps, the order of steps, etc. shown in the following embodiments are examples and do not limit the subject matter of the present disclosure. In addition, among the constituent elements in the following embodiments, the constituent elements recorded in the independent claims representing the highest concept are described as arbitrary constituent elements. In addition, in all embodiments, the various contents can also be combined.

[0051] (First embodiment)

[0052] Hereinafter, a first embodiment of the present disclosure will be described with reference to the drawings. In the following first embodiment, the same reference numerals are given to the same parts, and duplicate descriptions are omitted.

[0053] Figure 1 1 is a diagram showing the overall configuration of a model manufacturing system 100 in the first embodiment of the present disclosure. The model manufacturing system 100 includes a battery mounting device 1 and a server 2 (manufacturing device). The battery mounting device 1 and the server 2 are communicably connected via a network 4. The network 4 is, for example, the Internet.

[0054] The battery-carrying device 1 is, for example, an electric vehicle such as an electric car, an electric truck, an electric motorcycle, or an electric bicycle, which is equipped with a rechargeable battery 11 and moves with the power charged by the battery 11. The battery-carrying device 1 is not limited thereto, and may also be an electric mobile body such as an unmanned aerial vehicle, a ship, or a robot, which is equipped with a rechargeable battery 11 and moves with the power charged by the battery 11. In addition, the battery-carrying device 1 may also be a stationary power supply device that carries a battery 11 that stores power supplied from a solar power generation device and a power supply company, etc., and supplies the power charged by the battery 11 to a facility.

[0055] The battery-mounted device 1 periodically transmits log data indicating the state of the storage battery 11 to the server 2. The log data will be described in detail later.

[0056] The server 2 is, for example, a cloud server. The server 2 receives various information such as log data from the battery-mounted device 1. The server 2 generates a plurality of degradation degree estimation models (learned models) for estimating the degradation degree of the storage battery 11 based on the log data obtained from the battery-mounted device 1. The degradation degree of the storage battery 11 represents how much the storage battery 11 has deteriorated relative to the initial state. The server 2 outputs a degradation degree estimation model having the best estimation accuracy of the degradation degree among the plurality of degradation degree estimation models.

[0057] The following describes the configurations of the battery mounting device 1 and the server 2. The battery mounting device 1 includes a storage battery 11, a memory 12, a communication unit 13, an operation unit 14, and a control unit 15.

[0058] The storage battery 11 is, for example, a rechargeable and dischargeable secondary battery such as a lithium ion battery. The storage battery 11 discharges (supplies) the power stored in itself to various operating parts of the battery-mounted device 1. In addition, when the battery-mounted device 1 is a stationary power supply device, the storage battery 11 discharges (supplies) the power stored in itself to an external device via a power cable (not shown).

[0059] The memory 12 is a storage device capable of storing various information, such as a RAM (Random Access Memory), an SSD (Solid State Drive), or a flash memory.

[0060] The communication unit 13 is a communication interface circuit that transmits and receives various information to and from an external device such as the server 2 via the network 4 .

[0061] The operation unit 14 receives an operation of the battery mounting device 1. The operation unit 14 includes, for example, a liquid crystal display and a touch panel device.

[0062] The control unit 15 is, for example, a microcomputer including a processor, a nonvolatile memory such as a RAM (Random Access Memory) and a ROM (Read Only Memory), an input / output circuit, a timer circuit, and various measuring circuits. The various measuring circuits include measuring circuits for measuring the current, voltage, and temperature of the battery 11, respectively.

[0063] The control unit 15 charges the storage battery 11 with electric power supplied via a power cable (not shown).

[0064] The control unit 15 measures the state of the storage battery 11 periodically, for example, every 10 seconds, and stores log data indicating the measurement result in the memory 12. The log data includes, for example, the time when the state of the storage battery 11 was measured (hereinafter referred to as the measurement time) and information indicating the measurement result of the state of the storage battery 11 (hereinafter referred to as the state information). The state of the storage battery 11 includes the current, voltage, and temperature of the storage battery 11.

[0065] The control unit 15 periodically sends the log data stored in the memory 12 to the server 2 using the communication unit 13. The timing at which the control unit 15 sends the log data to the server 2 is not limited thereto. For example, the control unit 15 may send the log data stored in the memory 12 to the server 2 using the communication unit 13 at the timing at which the battery 11 is charged and discharged. Alternatively, the control unit 15 may send the log data stored in the memory 12 to the server 2 using the communication unit 13 at the timing at which a given number of charge and discharge operations are completed.

[0066] When the communication unit 13 receives the degradation degree estimation model for estimating the degradation degree of the storage battery 11 from the server 2, the control unit 15 stores the degradation degree estimation model in the memory 12. In addition, the control unit 15 uses the communication unit 13 to request the server 2 to send the degradation degree estimation model at the timing when the charging and discharging of the storage battery 11 are completed. In response to the request, the control unit 15 stores the degradation degree estimation model received by the communication unit 13 from the server 2 in the memory 12. The timing at which the control unit 15 requests the server 2 to send the degradation degree estimation model is not limited to this. For example, the control unit 15 may request the server 2 to send the degradation degree estimation model in response to the user's operation on the operation unit 14.

[0067] When the deterioration degree estimation model receives log data indicating the state of the storage battery 11 during charge and discharge, it outputs the deterioration degree of the storage battery 11 during charge and discharge. The charge and discharge period refers to the period from the start to the end of charging or discharging of the storage battery 11.

[0068] The control unit 15 estimates the degree of degradation of the storage battery 11 at the timing when the charge and discharge of the storage battery 11 are completed. The timing at which the control unit 15 estimates the degree of degradation of the storage battery 11 is not limited to this. For example, the control unit 15 may estimate the degree of degradation of the storage battery 11 in accordance with the operation of the operation unit 14 by the user.

[0069] Specifically, the control unit 15 obtains the degradation degree estimation model stored in the memory 12, and inputs the log data stored in the memory 12 during the most recent charge and discharge into the degradation degree estimation model. Thus, if the degradation degree estimation model outputs the degradation degree of the battery 11, the control unit 15 displays the degradation degree on the liquid crystal display provided in the operation unit 14.

[0070] The server 2 includes a memory 22 , a communication unit 23 , and a control unit 25 .

[0071] The communication unit 23 is a communication interface circuit that transmits and receives various information to and from the external device such as the battery-mounted device 1 via the network 4. For example, when the communication unit 23 receives log data from the battery-mounted device 1, it outputs the received log data to the control unit 25. In addition, under the control of the control unit 25, the communication unit 23 transmits the degradation degree estimation model with the best estimation accuracy of the degradation degree to the battery-mounted device 1.

[0072] The memory 22 is a storage device capable of storing various information, such as a RAM, a HDD (Hard Disk Drive), an SSD, or a flash memory. The memory 22 stores a control program executed by the control unit 25 .

[0073] The memory 22 stores the log data received by the communication unit 23 from the battery-mounted device 1. The memory 22 stores information related to the storage battery 11 (hereinafter referred to as storage battery information). The storage battery information includes a characteristic table showing the relationship between the SOC (State of Charge) of the storage battery 11 and the voltage of the storage battery 11, and the full charge capacity of the storage battery 11 in the initial state. SOC is an indicator representing the charging rate of the storage battery 11. SOC is represented by (remaining capacity [Ah] / full charge capacity [Ah]).

[0074] The control unit 25 is, for example, a CPU and functions as an acquisition unit 251 , a calculation unit 252 , a generation unit 253 (a first generation unit and a second generation unit), an evaluation unit 254 , and an output unit 255 by executing a control program stored in the memory 22 .

[0075] The acquisition unit 251 acquires the log data received by the communication unit 23 from the battery mounting device 1 , and stores the log data in the memory 22 .

[0076] Furthermore, in a model creation process described later, the acquisition unit 251 acquires a plurality of log data indicating the state of the storage battery 11 during charging or discharging (hereinafter referred to as first log data) from the memory 22 .

[0077] Specifically, the acquisition unit 251 refers to the current value of the storage battery 11 included in the log data to determine whether the storage battery 11 is being charged or discharged when the log data is acquired. By making this determination, the acquisition unit 251 acquires a plurality of first log data indicating the state of the storage battery 11 during charging or discharging.

[0078] For example, it is assumed that the current value of the storage battery 11 included in the log data is within a given range close to 0. In this case, the acquisition unit 251 determines that the storage battery 11 is not being charged or discharged when the log data is acquired, and that the storage battery 11 is in a rest state.

[0079] Assume that the current value of the storage battery 11 included in the log data is a current value during discharge outside the given range (for example, a negative value). In this case, the acquisition unit 251 determines whether the storage battery 11 is discharged when the log data is acquired. Assume that the current value of the storage battery 11 included in the log data is a current value during charge outside the given range (for example, a positive value). In this case, the acquisition unit 251 determines that the storage battery 11 is charged when the log data is acquired.

[0080] The acquisition unit 251 calculates the degradation degree (hereinafter referred to as the first degradation degree) of the storage battery 11 during the charge and discharge (hereinafter referred to as the first charge and discharge) corresponding to each acquired first log data by a degradation degree estimation method. The first charge and discharge refers to the charge or discharge of the storage battery 11 performed in the battery-mounted device 1 when the first log data is acquired.

[0081] In the present embodiment, the degradation degree of the storage battery 11 is represented by the SOH (State Of Health) of the storage battery 11. The SOH is represented by (full charge capacity [Ah] at the time of degradation (current) / full charge capacity [Ah] when the storage battery 11 is in the initial state). In addition, the acquisition unit 251 uses the first log data to calculate the first degradation degree by a two-point OCV estimation method widely known as a degradation degree estimation method.

[0082] Specifically, the acquisition unit 251 refers to the log data stored in the memory 22, and acquires the voltage of the storage battery 11 when the storage battery 11 is in a dormant state immediately before and immediately after the first charge and discharge, as the open circuit voltage of the storage battery 11 immediately before and immediately after the first charge and discharge. The so-called immediately before the charge or discharge means immediately before the charge or discharge of the storage battery 11 is started. The so-called immediately after the charge or discharge means immediately after the charge or discharge of the storage battery 11 is finished.

[0083] The acquisition unit 251 refers to the characteristic table stored in the memory 22, and determines the open circuit voltage of the storage battery 11 immediately before and immediately after the first charge and discharge as the SOC of the storage battery 11 at the start and end of the first charge and discharge. The acquisition unit 251 calculates the difference between the SOC of the storage battery 11 at the start of the first charge and discharge and the SOC of the storage battery 11 at the end of the first charge and discharge (hereinafter referred to as the SOC difference).

[0084] The acquisition unit 251 refers to the current value of the storage battery 11 included in the first log data to calculate the integrated value of the current of the storage battery 11 during the first charge and discharge. The acquisition unit 251 divides the calculated integrated value by the SOC difference (= integrated value of current / SOC difference) and further divides it by the full charge capacity of the storage battery 11 in the initial state stored in the memory 22. The calculation unit 252 calculates the result of the division (= integrated value of current / (SOC difference×full charge capacity of the storage battery 11 in the initial state)) as the first degree of degradation.

[0085] The calculation unit 252 calculates the reliability indicating the certainty of the first degradation degree corresponding to each first log data based on the content of the first charge and discharge corresponding to each first log data acquired by the acquisition unit 251. The first degradation degree corresponding to the first log data indicates the first degradation degree at the first charge and discharge corresponding to the first log data, calculated using the first log data.

[0086] Specifically, the calculation unit 252 calculates the reliability by multiplying the initial value of the reliability by a coefficient according to the content of the first charge and discharge.

[0087] Figure 2 This is a diagram showing an example of the relationship between the content of charge and discharge and the coefficient. For example, Figure 2 The coefficient includes the coefficient "ΔSOC" which is the difference between the SOC of the battery 11 at the start of the first charge / discharge and the SOC of the battery 11 at the end of the first charge / discharge. That is, the calculation unit 252 sets the reliability to a given initial value (e.g., 1), and then calculates the reliability by multiplying the SOC difference (e.g., 0.9) calculated when calculating the first degree of degradation, and calculates the result (e.g., 0.9) as the reliability.

[0088] Figure 2 It further indicates that the coefficient "0.7" (a coefficient less than 1) is determined in correspondence with the case where the time (hereinafter referred to as the rest time) of the storage battery 11 immediately before the first charge / discharge is less than 100 seconds ("rest time < 100 ms"). That is, when the rest time immediately before the first charge / discharge is less than 100 seconds (predetermined time), the calculation unit 252 calculates the reliability (for example, 0.9) calculated so far and multiplied by the coefficient "0.7" (for example, 0.63) as the reliability.

[0089] Figure 2 It further indicates that the coefficient "0.1" is determined in correspondence with the case where the first degree of degradation corresponding to each first log data is greater than 1.1 (a given upper limit value) ("SOH>1.1"). That is, the calculation unit 252 calculates the reliability calculated so far (for example, 0.63) by further multiplying the coefficient "0.1" (for example, 0.063) as the reliability when the first degree of degradation corresponding to each first log data is greater than 1.1.

[0090] Figure 2 It further indicates that the coefficient "0.1" is determined in correspondence with the case where the first degree of degradation corresponding to each first log data is less than 0.5 (a given lower limit value) ("SOH<0.5"). That is, when the first degree of degradation corresponding to each first log data is less than 0.5, the calculation unit 252 calculates the reliability calculated so far (for example, 0.63) by further multiplying the coefficient "0.1" (for example, 0.063) as the reliability.

[0091] The generation unit 253 generates a first degradation degree estimation model (first learned model) by performing machine learning on the relationship between the plurality of first degradation degrees calculated by the acquisition unit 251 and the plurality of first log data acquired by the acquisition unit 251 .

[0092] In addition, the generator 253 generates a second degradation degree estimation model (second learned model) by performing machine learning on the relationship between the first degradation degree whose reliability is greater than a given value among the plurality of first degradation degrees calculated by the acquirer 251 and the first log data corresponding to the first degradation degree. The first log data corresponding to the first degradation degree refers to the first log data used in calculating the first degradation degree. The given value is determined to be a value (e.g., 0.8) less than the initial value (e.g., 1) of the reliability. In other words, the initial value of the reliability is determined to be a value greater than the given value.

[0093] Figure 31 is a diagram showing an example of a feature quantity used as an explanatory variable for machine learning. Specifically, when generating the first degradation degree estimation model, the generating unit 253 calculates the voltage, current, and temperature of the battery 11 included in the first log data corresponding to the first degradation degree. Figure 3 The characteristic quantities shown.

[0094] Specifically, the generator 253 classifies the period during which the first charge and discharge is being performed into 10 periods according to the SOC of the battery 11 .

[0095] In addition, the generation unit 253 calculates the SOC of the battery 11 using, for example, the SOC of the battery 11 at the start of the first charge / discharge, the integrated value of the current of the battery 11 included in the first log data from the start of the first charge / discharge, and the full charge capacity of the battery 11 in the initial state. The SOC of the battery 11 at the start of the first charge / discharge may be calculated using the SOC calculated by the acquisition unit 251 when calculating the first degree of degradation.

[0096] The generator 253 calculates the average value, variance value, skewness, and kurtosis of the current, voltage, temperature, current difference, voltage difference, and temperature difference of the storage battery 11 in each classified period as feature amounts.

[0097] The current difference indicates the amount of change in the current of the storage battery 11. The generator 253 obtains the current value of the storage battery 11 from the first log data including the most recent measurement time that is past the measurement time included in the first log data. The generator 253 calculates the current difference by subtracting the obtained current value from the current value of the storage battery 11 included in the first log data.

[0098] The voltage difference indicates the amount of change in the voltage of the storage battery 11. The generator 253 obtains the voltage value of the storage battery 11 from the first log data including the most recent measurement time that is past the measurement time included in the first log data. The generator 253 calculates the voltage difference by subtracting the obtained voltage value from the voltage value of the storage battery 11 included in the first log data.

[0099] The temperature difference indicates the amount of change in the temperature of the storage battery 11. The generation unit 253 obtains the temperature value of the storage battery 11 from the first log data including the most recent measurement time that is past the measurement time included in the first log data. The generation unit 253 calculates the temperature difference by subtracting the obtained temperature value from the temperature value of the storage battery 11 included in the first log data.

[0100] For example, the first log data corresponding to the first degree of degradation is log data acquired when the SOC of the battery 11 is charged from 0% to 20%. In this case, the generator 253 calculates the average value "F011", the variance value "F012", the skewness "F013", and the kurtosis "F014" of the current of the battery 11 during the period from 0% to 10% of the SOC of the battery 11 as feature quantities.

[0101] Similarly, the generating unit 253 calculates as feature quantities the average value "F021", variance value "F022", skewness "F023" and kurtosis "F024" of the voltage of the battery 11 during the period from 0% to 10% of the SOC of the battery 11, the average value "F031", variance value "F032", skewness "F033" and kurtosis "F034" of the temperature, the average value "F041", variance value "F042", skewness "F043" and kurtosis "F044" of the current difference, the average value "F051", variance value "F052", skewness "F053" and kurtosis "F054" of the voltage difference, and the average value "F061", variance value "F062", skewness "F063" and kurtosis "F064" of the temperature difference.

[0102] The generator 253 also calculates the average value “F111”, variance value “F112”, skewness “F113”, and kurtosis “F114” of the current of the battery 11 during the period from 10% to 20% of the SOC of the battery 11, the average value “F121”, variance value “F122”, skewness “F123”, and kurtosis “F124” of the voltage, the average value “F131”, variance value “F132”, skewness “F133”, and kurtosis “F134” of the temperature. 133" and kurtosis "F134", the average value "F141", variance value "F142", skewness "F143" and kurtosis "F144" of the current difference, the average value "F151", variance value "F152", skewness "F153" and kurtosis "F154" of the voltage difference, and the average value "F161", variance value "F162", skewness "F163" and kurtosis "F164" of the temperature difference are used as feature quantities.

[0103] The generator 253 uses the feature amount calculated with reference to the first log data as an explanatory variable and the first degradation degree corresponding to the first log data as a target variable to perform machine learning using a given learning algorithm. Thus, the generator 253 generates a first degradation degree estimation model that outputs the degradation degree of the storage battery 11 during charging and discharging corresponding to the first log data when the first log data is input.

[0104] Similarly, the generator 253 calculates a feature value by referring to the first log data corresponding to the first degree of degradation whose reliability is greater than or equal to a given value. The generator 253 uses the feature value as an explanatory variable and the first degree of degradation as a target variable to perform machine learning using a given learning algorithm. Thus, the generator 253 generates a second degree of degradation estimation model that outputs the degree of degradation of the storage battery 11 during charging and discharging corresponding to the first log data when the first log data is input.

[0105] The evaluation unit 254 evaluates the estimation accuracy of the degree of degradation of the battery 11 in each of the first degradation degree estimation model and the second degradation degree estimation model generated by the generation unit 253 .

[0106] Specifically, the evaluation unit 254 obtains log data indicating the state of the storage battery 11 during the second charge (hereinafter referred to as the second log data) by referring to the log data stored in the memory 22. The second charge is a charge in which the storage battery 11 is charged when the temperature of the storage battery 11 is 20 degrees or more and within 30 degrees until the SOC of the storage battery 11 changes from 0% to 100%.

[0107] The evaluation unit 254 calculates the second degree of degradation, which is the degree of degradation of the storage battery 11 during the second charge, by the two-point OCV estimation method using the second log data, similarly to the acquisition unit 251 .

[0108] The evaluation unit 254 calculates the degree of deviation between the degradation degree of the storage battery 11 output (estimated) by inputting the second log data to each degradation degree estimation model and the second degradation degree, respectively. The degree of deviation is, for example, a root mean square error (RMSE) or a mean square error (MSE). However, the degree of deviation is not limited to this.

[0109] The evaluation unit 254 then evaluates the degradation degree estimation model with the smallest deviation between the first degradation degree estimation model and the second degradation degree estimation model as the degradation degree estimation model with the best estimation accuracy of the degradation degree of the battery 11 (hereinafter referred to as the best evaluation model). The evaluation unit 254 stores the best evaluation model in the memory 22 .

[0110] The output unit 255 transmits (outputs) the optimal evaluation model stored in the memory 22 to the battery mounting device 1 using the communication unit 23 .

[0111] When the communication unit 23 receives a request to transmit the degradation degree estimation model from the battery mounting device 1 , the output unit 255 replies (outputs) the optimal evaluation model stored in the memory 22 to the battery mounting device 1 using the communication unit 23 .

[0112] (Model making process)

[0113] Next, the model creation process performed in the server 2 is described. The model creation process is a process of generating a plurality of degradation degree estimation models for estimating the degradation degree of the storage battery 11 based on the log data obtained from the battery-mounted device 1, and outputting a degradation degree estimation model having the best estimation accuracy of the degradation degree among the plurality of degradation degree estimation models.

[0114] Figure 4 2 is a flowchart showing an example of the model making process. The control unit 25 periodically executes the model making process at a given cycle such as once a day. The timing of the control unit 25 executing the model making process is not limited to this. For example, whenever the communication unit 23 receives log data indicating that the storage battery 11 is in a rest state just after charging or discharging from the battery mounting device 1, that is, whenever charging or discharging is performed in the battery mounting device 1, the control unit 25 can execute the model making process.

[0115] First, in step S1 , the acquisition unit 251 acquires a plurality of first log data indicating the state of the storage battery 11 during charging or discharging from the memory 22 .

[0116] Next, in step S2, the acquisition unit 251 calculates the first degradation degree, which is the degradation degree of the storage battery 11 at the first charge and discharge time corresponding to each first log data acquired in step S1, by using the degradation degree estimation method. Thus, the acquisition unit 251 calculates a plurality of first degradation degrees.

[0117] Next, in step S3 , the calculation unit 252 calculates the reliability of the first degree of degradation corresponding to each first log data based on the content of the first charge and discharge corresponding to each first log data acquired in step S1 .

[0118] Next, in step S4 , the generator 253 generates a first degradation degree estimation model by performing machine learning on the relationship between the plurality of first degradation degrees calculated in step S2 and the plurality of first log data acquired in step S1 .

[0119] Next, in step S5, the generator 253 generates a second degradation degree estimation model by performing machine learning on the relationship between the first degradation degree having a reliability greater than a given value calculated in step S3 among the plurality of first degradation degrees calculated in step S2 and the first log data corresponding to the first degradation degree.

[0120] Next, in step S6, the evaluation unit 254 evaluates the estimation accuracy of the degradation degree of the battery 11 of each of the first degradation degree estimation model and the second degradation degree estimation model generated in step S4 and step S5. The evaluation unit 254 stores the best evaluation model evaluated as having the best estimation accuracy among the first degradation degree estimation model and the second degradation degree estimation model in the memory 22.

[0121] Next, in step S7 , the output unit 255 transmits the optimal evaluation model stored in the memory 22 to the battery mounting device 1 using the communication unit 23 .

[0122] As described above, in the structure of the first embodiment, a plurality of first log data indicating the state of the battery during charging or discharging is obtained from the battery mounting device 1. Therefore, a plurality of first log data required for machine learning for generating the first degradation degree estimation model and the second degradation degree estimation model can be obtained more quickly than in the case where a plurality of charging and discharging tests are performed in a test environment. Thus, this structure can quickly generate the first degradation degree estimation model and the second degradation degree estimation model.

[0123] In addition, in the configuration of the first embodiment, the second degradation degree estimation model is generated by performing machine learning on the relationship between the first degradation degree having a reliability greater than a given value and the first log data corresponding to the first degradation degree. Therefore, the present configuration can generate the second degradation degree estimation model that is considered to estimate the degradation degree of the storage battery 11 more accurately than the first degradation degree estimation model that may have performed machine learning on the relationship between the first degradation degree having a reliability less than a given value and the first log data used for calculating the first degradation degree.

[0124] However, even the second degradation degree estimation model generated in this way does not necessarily have a higher estimation accuracy of the degradation degree of the storage battery 11 than the first degradation degree estimation model. Therefore, in the structure of the first embodiment, the estimation accuracy of the degradation degree of the storage battery 11 of each of the first degradation degree estimation model and the second degradation degree estimation model is evaluated, and the best evaluation model evaluated as having the best estimation accuracy is sent to the battery mounting device 1. Therefore, this structure can send the degradation degree estimation model that can estimate the degradation degree of the storage battery 11 with good accuracy to the battery mounting device 1.

[0125] (Second embodiment)

[0126] Next, a second embodiment of the present disclosure is described. In the first embodiment, an example is described in which the server 2 generates a first degradation degree estimation model and a second degradation degree estimation model, and outputs the best evaluation model evaluated as having the best estimation accuracy of the degradation degree among the first degradation degree estimation model and the second degradation degree estimation model.

[0127] In the second embodiment, the server 2 further generates a third degradation degree estimation model (third learned model) different from the first degradation degree estimation model and the second degradation degree estimation model using the first log data. Then, the estimation accuracy of the degradation degree of each of the first degradation degree estimation model, the second degradation degree estimation model, and the third degradation degree estimation model is evaluated, and the best evaluation model evaluated as having the best estimation accuracy of the degradation degree is output. In addition, in the following description, the same reference numerals as those in the first embodiment are attached to the same components as those in the first embodiment, and their description is omitted.

[0128] In the second embodiment, the calculation unit 252 further extracts step S2 ( Figure 4 ) is a first degradation degree that is within a given deviation range among multiple first degradation degrees calculated in .

[0129] Figure 5 This is a diagram showing an example of a process of extracting the first degree of degradation within a given deviation range. Figure 5 The horizontal axis represents the time from when the battery 11 is in the initial state to step S2 ( Figure 4 ) is the square root of the time from the time point at which the first charge and discharge corresponding to each first degradation degree calculated in ) is completed. The first charge and discharge corresponding to the first degradation degree means the first charge and discharge corresponding to the first log data used in calculating the first degradation degree. Figure 5 The vertical axis of represents the degree of degradation of the storage battery 11. Reference numerals 91 to 96 represent step S2 ( Figure 4 ) is an example of multiple first degradation degrees calculated in ).

[0130] Specifically, Figure 5 As shown in FIG. 1 , the calculation unit 252 performs a linear regression from the time point when the storage battery 11 is in the initial state to step S2 ( Figure 4 ) are used as the explanatory variable, and the plurality of first degradation degrees 91 to 96 are used as the target variable.

[0131] The calculation unit 252 extracts step S2 ( Figure 4 ) among the multiple first degradation degrees 91 to 96 calculated in the above method, the first degradation degrees 91, 93, 95, and 96 whose distance to the regression line 80 obtained by linear regression is less than the given distance 89 are regarded as the first degradation degrees within the given deviation range.

[0132] The generation unit 253 further generates a third degradation degree estimation model by performing machine learning on the relationship between the first degradation degree within a given deviation range extracted by the calculation unit 252 and the first log data corresponding to the first degradation degree, similarly to the first degradation degree estimation model and the second degradation degree estimation model.

[0133] In step S6 ( Figure 4 ), further, similarly to the first degradation degree estimation model and the second degradation degree estimation model, the estimation accuracy of the degradation degree of the storage battery 11 in the third degradation degree estimation model is evaluated. The evaluation unit 254 stores the best evaluation model evaluated as having the best estimation accuracy of the degradation degree of the storage battery 11 among the first degradation degree estimation model, the second degradation degree estimation model, and the third degradation degree estimation model in the memory 22.

[0134] The output unit 255 performs the following steps in step S7 ( Figure 4 ) in which the optimal evaluation model stored in the memory 22 is sent to the battery mounting device 1 using the communication unit 23.

[0135] In the configuration of the second embodiment, the third degradation degree estimation model is generated by performing machine learning on the relationship between the first degradation degree within a given deviation range and the first log data corresponding to the first degradation degree. Therefore, the present configuration can generate the third degradation degree estimation model that estimates the degradation degree of the storage battery 11 more accurately than the first degradation degree estimation model that is considered to have performed machine learning on the relationship between the first degradation degree outside the deviation range and the first log data corresponding to the first degradation degree.

[0136] However, even the third degradation degree estimation model generated in this way does not necessarily have a higher estimation accuracy of the battery degradation degree than the first degradation degree estimation model and the second degradation degree estimation model. Therefore, in the structure of the second embodiment, the estimation accuracy of the degradation degree of the storage battery 11 of the third degradation degree estimation model is further evaluated, and the best evaluation model evaluated as having the best estimation accuracy among the first degradation degree estimation model, the second degradation degree estimation model, and the third degradation degree estimation model is sent to the battery mounting device 1. Therefore, this structure can send the degradation degree estimation model that can estimate the degradation degree of the storage battery 11 with good accuracy to the battery mounting device 1.

[0137] (Third embodiment)

[0138] Next, the third embodiment of the present disclosure is described. In the second embodiment, the following example is described: in the server 2, a first degradation degree estimation model, a second degradation degree estimation model, and a third degradation degree estimation model are generated, and the best evaluation model evaluated as having the best estimation accuracy of the degradation degree is output among the first degradation degree estimation model, the second degradation degree estimation model, and the third degradation degree estimation model.

[0139] In the third embodiment, the server 2 further generates a fourth degradation degree estimation model (fourth learned model) different from the first degradation degree estimation model, the second degradation degree estimation model, and the third degradation degree estimation model using the first log data. Then, the estimation accuracy of the degradation degree of each of the first degradation degree estimation model, the second degradation degree estimation model, the third degradation degree estimation model, and the fourth degradation degree estimation model is evaluated, and the best evaluation model evaluated as having the best estimation accuracy of the degradation degree is output. In addition, in the following description, the same reference numerals as those in the second embodiment are attached to the same components as those in the second embodiment, and their description is omitted.

[0140] In the third embodiment, the generation unit 253 further generates, similarly to the first degradation degree estimation model and the second degradation degree estimation model, a first degradation degree estimation model for the first degradation degree within the given deviation range extracted by the calculation unit 252 in step S3 ( Figure 4 ) is a fourth degradation degree estimation model that performs machine learning on the relationship between the first degradation degree whose reliability calculated in the method is greater than a given value and the first log data corresponding to the first degradation degree.

[0141] In step S6 ( Figure 4 ), the estimation accuracy of the degradation degree of the battery 11 in the fourth degradation degree estimation model is further evaluated in the same manner as the first degradation degree estimation model and the second degradation degree estimation model. The evaluation unit 254 stores the best evaluation model evaluated as having the best estimation accuracy of the degradation degree of the battery 11 among the first degradation degree estimation model, the second degradation degree estimation model, the third degradation degree estimation model, and the fourth degradation degree estimation model in the memory 22.

[0142] The output unit 255 performs the following steps in step S7 ( Figure 4 ) in which the optimal evaluation model stored in the memory 22 is sent to the battery mounting device 1 using the communication unit 23.

[0143] In the configuration of the third embodiment, a fourth degradation degree estimation model is generated by performing machine learning on the relationship between the first degradation degree having a reliability of a given value or more among the first degradation degrees within a given deviation range and the first log data corresponding to the first degradation degree. Therefore, this configuration can generate a fourth degradation degree estimation model that estimates the degradation degree of the storage battery 11 with higher accuracy than the third degradation degree estimation model that is considered to have performed machine learning on the relationship between the first degradation degree having a reliability of less than a given value among the first degradation degrees within the deviation range and the first log data used for calculating the first degradation degree.

[0144] However, even the fourth degradation degree estimation model generated in this way does not necessarily have a higher estimation accuracy of the degradation degree of the storage battery 11 than the first degradation degree estimation model, the second degradation degree estimation model, and the third degradation degree estimation model. Therefore, in the structure of the third embodiment, the estimation accuracy of the degradation degree of the storage battery 11 of the fourth degradation degree estimation model is further evaluated, and the degradation degree estimation model evaluated as having the best estimation accuracy among the first degradation degree estimation model, the second degradation degree estimation model, the third degradation degree estimation model, and the fourth degradation degree estimation model is sent to the battery mounting device 1. Therefore, this structure can output a degradation degree estimation model that can estimate the degradation degree of the storage battery 11 with good accuracy.

[0145] In the configurations of the first embodiment, the second embodiment, and the third embodiment described above, an example is described in which the acquisition unit 251 performs the following operations in step S1 ( Figure 4 ), a plurality of first log data are obtained from the memory 22, and in step S2 ( Figure 4 ), a plurality of first degradation degrees are calculated by using a degradation degree estimation method using the plurality of first log data.

[0146] However, instead, the acquisition unit 251 may acquire a plurality of first log data from an external device different from the server 2 using the communication unit 23 , and acquire a plurality of first degradation degrees calculated by a degradation degree estimation method using the plurality of first log data.

[0147] Similarly, in step S6 ( Figure 4 ), the evaluation unit 254 uses the communication unit 23 to obtain, from an external device different from the server 2, second log data indicating the state of the storage battery 11 during the second charge and a second degradation degree calculated by a degradation degree estimation method using the second log data.

[0148] The technology involved in the present disclosure can quickly output the learned model that can accurately estimate the degradation degree of the battery to a device equipped with a chargeable and dischargeable battery such as an electric vehicle, and is therefore useful in accurately displaying the degradation degree of the battery in the device.

Claims

1. A manufacturing method, in a manufacturing device using a learned model for estimating the degree of degradation of a chargeable and dischargeable battery, obtaining a plurality of first log data indicating the state of the battery during charging or discharging obtained from a device equipped with the battery and a plurality of first degradation degrees indicating the degree of degradation of the battery calculated by a degradation degree estimation method using each of the first log data, Based on the contents of the charge or discharge corresponding to each first log data, namely, the first charge and discharge, a reliability indicating the certainty of the first degree of degradation corresponding to each first log data is calculated. generating a first learned model by performing machine learning on the relationship between the plurality of first degradation degrees and the plurality of first log data, generating a second learned model by performing machine learning on the relationship between a first degradation degree having a reliability greater than or equal to a given value among the plurality of first degradation degrees and first log data corresponding to the first degradation degree, evaluating the estimation accuracy of the battery degradation degree in each of the first learned model and the second learned model, The learned model evaluated as having the best estimation accuracy among the first learned model and the second learned model is output.

2. The manufacturing method according to claim 1, wherein: further extracting a first degradation degree that is within a given deviation range from among the plurality of first degradation degrees, Further, a third learned model is generated by performing machine learning on the relationship between the extracted first degradation degree and the first log data corresponding to the first degradation degree. In the evaluation, the estimation accuracy in the third learned model is further evaluated. In the output, a learned model evaluated as having the best estimation accuracy among the first learned model, the second learned model, and the third learned model is output.

3. The manufacturing method according to claim 2, wherein: A fourth learned model is generated by further performing machine learning on the relationship between the first degradation degree whose reliability is greater than or equal to the given value among the extracted first degradation degrees and the first log data corresponding to the first degradation degree, In the evaluation, the estimation accuracy in the fourth learned model is further evaluated. In the output, a learned model evaluated as having the best estimation accuracy among the first learned model, the second learned model, the third learned model, and the fourth learned model is output.

4. The production method according to any one of claims 1 to 3, wherein In the evaluation, obtaining second log data indicating the state of the battery during a second charge in which the SOC of the battery is charged from 0% to 100% when the temperature of the battery is 20 degrees or more and 30 degrees or less, obtained from a device on which the battery is mounted, and a second degradation degree indicating the degree of degradation of the battery calculated by the degradation degree estimation method using the second log data, calculating, for each of the plurality of learned models that are the objects of the evaluation, a degree of deviation between the degradation degree of the battery estimated by inputting the second log data into each learned model and the second degradation degree, The learned model having the smallest degree of deviation among the plurality of learned models is evaluated as the learned model having the best estimation accuracy.

5. The manufacturing method according to claim 1, wherein: In calculating the reliability, a result obtained by multiplying an initial value of the reliability that is larger than the given value by a coefficient corresponding to the content of the first charge and discharge is calculated as the reliability.

6. The manufacturing method according to claim 5, wherein: The coefficient is determined as a difference between the SOC of the battery at the start of the first charge / discharge and the SOC of the battery at the end of the first charge / discharge.

7. The manufacturing method according to claim 5 or 6, wherein: In calculating the reliability, when the time during which the battery is in a rest state immediately before the first charge / discharge is less than a predetermined time, a coefficient less than 1 is multiplied.

8. The manufacturing method according to claim 5 or 6, wherein: In calculating the reliability, when the first degree of degradation corresponding to each first log data is greater than a given upper limit value or less than a given lower limit value, a coefficient less than 1 is multiplied.

9. The manufacturing method according to claim 7, wherein: In calculating the reliability, when the first degree of degradation corresponding to each first log data is greater than a given upper limit value or less than a given lower limit value, a coefficient less than 1 is multiplied.

10. The manufacturing method according to claim 1, wherein: In the above-mentioned degradation degree estimation method, obtaining the voltage of the battery when the battery is in a rest state immediately before and immediately after the first charge and discharge as the open circuit voltage of the battery immediately before and immediately after the first charge and discharge, respectively; referring to information indicating a relationship between the SOC of the battery and the open circuit voltage of the battery, determining the open circuit voltage of the battery immediately before and immediately after the first charge and discharge as the SOC of the battery at the start and end of the first charge and discharge, calculating a difference in the SOC of the battery at the start and end of the first charge / discharge operation, using each first log data to calculate a cumulative value of the current of the battery during the first charge and discharge; A result obtained by dividing the difference by the integrated value is further divided by the full charge capacity of the battery in an initial state, and the result is calculated as the degree of degradation of the battery.

11. The manufacturing method according to claim 4, wherein: In the above-mentioned degradation degree estimation method, immediately before and immediately after the second charging is performed, respectively, obtaining the voltage of the battery when the battery is in a rest state as the open circuit voltage of the battery immediately before and immediately after the second charging is performed, referring to information indicating a relationship between the SOC of the battery and the open circuit voltage of the battery, determining the open circuit voltage of the battery immediately before and immediately after the second charging as the SOC of the battery at the start and end of the second charging, calculating a difference in the SOC of the battery at the start and end of the second charging, using the second log data to calculate a cumulative value of the current of the battery during the second charge; A result obtained by dividing the difference by the integrated value is further divided by the full charge capacity of the battery in an initial state, and the result is calculated as the degree of degradation of the battery.

12. The manufacturing method according to claim 2 or 3, wherein: In extracting the first degradation degree within the deviation range, A linear regression is performed using the square root of the time from the time when the battery is in the initial state to the time when the first charge and discharge is completed as an explanatory variable and each first degree of degradation as a target variable, A first degradation degree whose distance to a regression line obtained by the linear regression is less than or equal to a given distance is extracted from among the plurality of first degradation degrees as a first degradation degree within the deviation range.

13. A manufacturing device, comprising: an acquisition unit that acquires a plurality of first log data indicating a state of the battery during charging or discharging, obtained from a device on which the battery is mounted, and a plurality of first degradation degrees indicating a degree of degradation of the battery calculated by a degradation degree estimation method using each of the first log data; a calculation unit that calculates a reliability indicating the certainty of a first degree of degradation corresponding to each first log data based on the content of the charge or discharge corresponding to each first log data, that is, the first charge and discharge; a first generating unit for generating a first learned model by performing machine learning on the relationship between the plurality of first degradation degrees and the plurality of first log data; a second generation unit that generates a second learned model by performing machine learning on a relationship between a first degradation degree having a reliability greater than or equal to a given value among the plurality of first degradation degrees and first log data corresponding to the first degradation degree; an evaluation unit that evaluates the estimation accuracy of the degree of degradation of the battery in each of the first learned model and the second learned model; and An output unit outputs a learned model evaluated as having the best estimation accuracy among the first learned model and the second learned model.

14. A program for a manufacturing device for estimating a learned model for the degree of degradation of a chargeable and dischargeable battery, the program causing the manufacturing device to execute the following processing: obtaining a plurality of first log data indicating the state of the battery during charging or discharging obtained from a device equipped with the battery and a plurality of first degradation degrees indicating the degree of degradation of the battery calculated by a degradation degree estimation method using each of the first log data, Based on the contents of the charge or discharge corresponding to each first log data, namely, the first charge and discharge, a reliability indicating the certainty of the first degree of degradation corresponding to each first log data is calculated. generating a first learned model by performing machine learning on the relationship between the plurality of first degradation degrees and the plurality of first log data, generating a second learned model by performing machine learning on the relationship between a first degradation degree having a reliability greater than or equal to a given value among the plurality of first degradation degrees and first log data corresponding to the first degradation degree, evaluating the estimation accuracy of the battery degradation degree in each of the first learned model and the second learned model, The learned model evaluated as having the best estimation accuracy among the first learned model and the second learned model is output.

Images