Information processing method, information processing device, and information processing program

Abstract

This information processing method comprises: acquiring a first use history of a first battery; acquiring a first predicted deterioration amount of the first battery by inputting the first use history into a first mathematical model; and outputting a second predicted deterioration amount of the first battery by inputting the first predicted deterioration amount and a first feature amount that is calculated from the first use history into a machine learning model.

Description

Information processing method, information processing device, and information processing program

[0001] The present disclosure relates to a technique for predicting the degree of deterioration of a battery.

[0002] For example, Patent Literature 1 discloses a battery management device that detects a predetermined state of a storage battery and estimates the predetermined state of the storage battery based on operation history information that indicates the history of operation of the storage battery, in accordance with a basic model for estimating the internal state of the storage battery and a degradation model that imparts a degradation function of the storage battery to the basic model. If the detected predetermined state and the estimated predetermined state show a deviation of a certain amount or more, the battery management device estimates the degradation factor of the storage battery and changes the degradation model based on the estimated degradation factor.

[0003] However, in the above-mentioned conventional technology, when the degradation factors change, it is necessary to change the degradation model. Therefore, when predicting the degradation of a battery used under conditions outside the range for which the degradation model was trained, it is difficult to accurately predict the degradation of the battery without changing the degradation model, and further improvement is needed.

[0004] Patent No. 6246539

[0005] The present disclosure has been made to solve the above problem, and aims to provide a technology that can improve the accuracy of predicting the deterioration of a battery used under conditions outside the range learned by a machine learning model.

[0006] An information processing method according to one aspect of the present disclosure is an information processing method executed by a computer, and includes acquiring a first usage history of a first battery, acquiring a first predicted deterioration amount of the first battery by inputting the first usage history into a first mathematical model, and outputting a second predicted deterioration amount of the first battery by inputting a first feature calculated from the first usage history and the first predicted deterioration amount into a machine learning model.

[0007] According to the present disclosure, it is possible to improve the accuracy of predicting the deterioration of a battery used under conditions outside the range learned by the machine learning model.

[0008] FIG. 1 is a diagram showing a configuration of an information processing system according to the first embodiment. FIG. 2 is a diagram showing an example of a usage history in the first embodiment. FIG. 3 is a diagram showing an example of explanatory variables and target variables in the first embodiment. FIG. 4 is a flowchart for explaining a prediction process by the information processing device in the first embodiment. FIG. 5 is a flowchart for explaining a learning process by the information processing device in the first embodiment. FIG. 6 is a diagram showing a configuration of an information processing system according to the second embodiment. FIG. 7 is a flowchart for explaining a learning process by the information processing device in the second embodiment.

[0009] (Knowledge forming the basis of the present disclosure) When predicting the degradation of lithium-ion batteries, machine learning models such as decision trees or neural networks can be expected to have high prediction accuracy, but there is a risk that the prediction accuracy will decrease for batteries used under conditions outside the learned range.

[0010] When predicting battery degradation using a machine learning model, the prediction accuracy is high for interpolated data within the range that the machine learning model has learned, but the prediction accuracy is low for extrapolated data outside the range that the machine learning model has learned. Testing all battery products under all usage conditions is difficult in terms of time and cost.

[0011] In the above-mentioned conventional technology, when predicting the deterioration of a battery used under conditions outside the range for which the deterioration model was learned, it was difficult to accurately predict the deterioration of the battery without changing the deterioration model.

[0012] In order to solve the above problems, the following techniques are disclosed.

[0013] (1) An information processing method according to one aspect of the present disclosure is an information processing method executed by a computer, and includes: acquiring a first usage history of a first battery; acquiring a first predicted deterioration amount of the first battery by inputting the first usage history into a first mathematical model; and outputting a second predicted deterioration amount of the first battery by inputting a first feature calculated from the first usage history and the first predicted deterioration amount into a machine learning model.

[0014] According to this configuration, a first predicted deterioration amount is obtained using a first mathematical model constructed using a theoretical formula derived from the chemical deterioration characteristics of the battery, and the obtained first predicted deterioration amount is input into a machine learning model to calculate a second predicted deterioration amount of the first battery, thereby improving the accuracy of predicting deterioration of a battery used under conditions outside the range learned by the machine learning model.

[0015] (2) In the information processing method described above in (1), the first mathematical model may include a theoretical formula for calculating a predicted amount of deterioration due to storage deterioration and a theoretical formula for calculating a predicted amount of deterioration due to cycle deterioration.

[0016] According to this configuration, it is possible to calculate the first predicted deterioration amount due to storage deterioration and cycle deterioration of the first battery.

[0017] (3) In the information processing method described in (1) or (2) above, the machine learning model may be trained using, as explanatory variables, features calculated from the battery's usage history and a predicted amount of deterioration of the battery calculated by inputting the usage history into the first mathematical model, and using, as a target variable, the actually measured amount of deterioration of the battery.

[0018] According to this configuration, when a feature calculated from the battery's usage history and a predicted amount of battery deterioration calculated by inputting the usage history into the first mathematical model are input, the machine learning model can be trained to output the predicted amount of battery deterioration.

[0019] (4) In the information processing method described in any one of (1) to (3) above, the method may further include obtaining a second usage history of a second battery different from the first battery, and creating the first mathematical model based on the second usage history.

[0020] According to this configuration, the first mathematical model can be created based on the second usage history of the second battery, which is different from the first battery.

[0021] (5) In the information processing method described in (4) above, the method may further include calculating a third predicted deterioration amount of the second battery by inputting the second usage history into the first mathematical model, and training the machine learning model using the second feature calculated from the second usage history and the third predicted deterioration amount as explanatory variables and the actually measured deterioration amount of the second battery as a dependent variable.

[0022] According to this configuration, when the second feature calculated from the second usage history and the third predicted deterioration amount calculated by inputting the second usage history into the first mathematical model are input, the machine learning model can be trained to output the predicted deterioration amount of the second battery.

[0023] (6) In the information processing method described in any one of (1) to (3) above, the method may further include acquiring a second usage history of a second battery different from the first battery, acquiring a third usage history of a third battery different from the first battery and the second battery, creating a second mathematical model based on the second usage history, calculating a fourth predicted deterioration amount of the third battery by inputting the third usage history into the second mathematical model, and training the machine learning model using a third feature calculated from the third usage history and the fourth predicted deterioration amount as explanatory variables and an actually measured measured deterioration amount of the third battery as a dependent variable.

[0024] According to this configuration, the third feature amount calculated from the third usage history of the third battery and the fourth predicted deterioration amount of the third battery calculated by inputting the third usage history into the second mathematical model created based on the second usage history of the second battery are used as explanatory variables of the machine learning model. Therefore, since the battery usage history used to create the mathematical model differs from the battery usage history input into the mathematical model, it is possible to create a machine learning model that predicts with higher accuracy the deterioration of a battery used under conditions outside the learned range.

[0025] (7) In the information processing method described in (6) above, the method may further include creating a third mathematical model based on the third usage history, calculating a fifth predicted deterioration amount of the second battery by inputting the second usage history into the third mathematical model, and training the machine learning model using the second feature calculated from the second usage history and the fifth predicted deterioration amount as explanatory variables and the actually measured measured deterioration amount of the second battery as a dependent variable.

[0026] According to this configuration, the second feature amount calculated from the second usage history of the second battery and the fifth predicted deterioration amount of the second battery calculated by inputting the second usage history into a third mathematical model created based on the third usage history of the third battery are used as explanatory variables of the machine learning model. Therefore, the battery usage history used to create the mathematical model differs from the battery usage history input into the mathematical model, making it possible to create a machine learning model that predicts with higher accuracy the deterioration of a battery used under conditions outside the learned range.

[0027] (8) The information processing method according to (6) or (7) above may further include creating the first mathematical model based on the second usage history and the third usage history.

[0028] According to this configuration, the first mathematical model can be created based on the second usage history of the second battery and the third usage history of the third battery.

[0029] (9) In the information processing method described in any one of (1) to (8) above, the first characteristic amount may include a cumulative elapsed time of the first battery used under each of a plurality of usage conditions represented by a combination of a predetermined range of SOC of the first battery, a predetermined range of current rate, and a predetermined range of temperature.

[0030] With this configuration, the second predicted deterioration amount can be calculated according to various conditions of use of the first battery.

[0031] Furthermore, the present disclosure can be realized not only as an information processing method that executes the characteristic processes described above, but also as an information processing device having a characteristic configuration corresponding to the characteristic processes executed by the information processing method. Furthermore, the present disclosure can also be realized as a computer program that causes a computer to execute the characteristic processes included in such an information processing method. Therefore, the same effects as those of the above information processing method can also be achieved in the following other aspects.

[0032] (10) Another aspect of the present disclosure is an information processing device including a processor, wherein the processor acquires a first usage history of a first battery, acquires a first predicted deterioration amount of the first battery by inputting the first usage history into a first mathematical model, and outputs a second predicted deterioration amount of the first battery by inputting a first feature calculated from the first usage history and the first predicted deterioration amount into a machine learning model.

[0033] (11) An information processing program according to another aspect of the present disclosure causes a computer to function as follows: acquiring a first usage history of a first battery; acquiring a first predicted deterioration amount of the first battery by inputting the first usage history into a first mathematical model; and outputting a second predicted deterioration amount of the first battery by inputting a first feature calculated from the first usage history and the first predicted deterioration amount into a machine learning model.

[0034] A non-transitory computer-readable recording medium according to another aspect of the present disclosure records the information processing program described in (11) above.

[0035] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. Note that each of the embodiments described below represents a specific example of the present disclosure. The numerical values, shapes, components, steps, and order of steps shown in the following embodiments are merely examples and are not intended to limit the present disclosure. Furthermore, among the components in the following embodiments, components that are not described in the independent claims that represent the highest concept are described as optional components. Furthermore, in all embodiments, the respective contents can be combined.

[0036] First Embodiment FIG. 1 is a diagram showing the configuration of an information processing system 1 according to a first embodiment.

[0037] The information processing system 1 includes a first battery 11 , a second battery 12 , an information processing device 20 , and a terminal 30 .

[0038] The terminal 30 is, for example, a personal computer, a smartphone, or a tablet computer, and displays information to be presented to a user. The terminal 30 is connected to the information processing device 20 via a network so that they can communicate with each other. The network is a local area network or a wide area network. The terminal 30 may be connected to the information processing device 20 via a wired or wireless connection so that they can communicate with each other.

[0039] The first battery 11 is a secondary battery for which degradation is predicted, and stores power by charging and supplies power by discharging. The secondary battery is, for example, a lithium-ion battery. The first battery 11 is installed as a power source in various devices. The first battery 11 is connected to the information processing device 20 via a network so that they can communicate with each other.

[0040] The first battery 11 transmits a first usage history to the information processing device 20. The first usage history includes a date and time, a current value, a voltage value, a capacity, a temperature, an SOC (State of Charge), and an SOH (State of Health). The first battery 11 periodically measures the current value, the voltage value, the capacity, the temperature, the SOC, and the SOH.

[0041] The current value, voltage value, and capacity of the first battery 11 are measured by a measuring device (not shown) provided in the first battery 11. The temperature of the first battery 11 is measured by a temperature sensor (not shown) provided in the first battery 11. The SOC is an index representing the charging rate of the first battery 11. The SOC of the first battery 11 is expressed by (remaining capacity [Ah] / fully charged capacity [Ah]) * 100.

[0042] The SOH is an index representing the health of the first battery 11. The SOH of the first battery 11 is expressed as (full charge capacity [Ah] at the time of deterioration (at a specified time) / initial full charge capacity [Ah]) * 100. The SOH represents how much the battery has deteriorated from the time the battery was first used until a specified time. The SOH is an example of the amount of deterioration, and is also called the degree of deterioration.

[0043] Note that the frequency of measuring the current value, voltage value, capacity, temperature, and SOC is different from the frequency of measuring the SOH. The frequency of measuring the current value, voltage value, capacity, temperature, and SOC is higher than the frequency of measuring the SOH. For example, the current value, voltage value, capacity, temperature, and SOC are measured every second, every 10 seconds, or every minute, and the SOH is measured once a month. Therefore, from the first date and time when the SOH is measured to the second date and time when the SOH is next measured, the SOH measured on the first date and time is included in the first usage history.

[0044] The second battery 12 is a battery different from the first battery 11. The second battery 12 is a secondary battery used for creating the first mathematical model and training the machine learning model, and stores power by charging and supplies power by discharging. The second battery 12 is connected to the information processing device 20 via a network so that they can communicate with each other.

[0045] The first battery 11 and the second battery 12 may be batteries manufactured by different manufacturers. The first battery 11 and the second battery 12 are preferably the same type of secondary battery (e.g., lithium ion battery). The first battery 11 and the second battery 12 may have different capacities.

[0046] The first battery 11 and the second battery 12 may be different types of lithium ion batteries. Lithium ion batteries are classified into several types depending on the material used in the positive electrode. The first battery 11 and the second battery 12 may be any of cobalt-based lithium ion batteries, manganese-based lithium ion batteries, nickel-based lithium ion batteries, NCA-based lithium ion batteries, iron phosphate-based lithium ion batteries, and ternary lithium ion batteries.

[0047] Furthermore, the secondary battery is not limited to a lithium ion battery, but may be a lead storage battery, a nickel-metal hydride battery, an all-solid-state battery, or the like.

[0048] The second battery 12 transmits a second usage history to the information processing device 20. The second usage history includes the date and time, current value, voltage value, capacity, temperature, SOC, and SOH. The second battery 12 periodically measures the current value, voltage value, capacity, temperature, SOC, and SOH. The content of the second usage history is the same as the content of the first usage history.

[0049] The information processing device 20 includes a first usage history acquisition unit 101, a first deterioration prediction unit 102, a first feature calculation unit 103, a second deterioration prediction unit 104, an output unit 105, a second usage history acquisition unit 111, a first mathematical model creation unit 112, a third deterioration prediction unit 113, a second feature calculation unit 114, a learning unit 115, a mathematical model storage unit 201, and a machine learning model storage unit 202.

[0050] The information processing device 20 includes at least a computer system including, for example, a control program, a processing circuit such as a processor or logic circuit that executes the control program, and a recording device such as an internal memory or an accessible external memory that stores the control program. Note that the information processing device 20 may be realized, for example, by hardware implementation using the processing circuit, or by execution of a software program stored in the memory by the processing circuit or distributed from an external server, or by a combination of these hardware and software implementations.

[0051] The information processing device 20 may be a server or an edge computer. The configuration of the information processing device 20 may be implemented in the terminal 30. The configuration of the information processing device 20 may be distributed among the terminal 30 and the server.

[0052] The information processing device 20 performs a prediction process to predict the deterioration amount of the first battery 11 and a learning process to create a first mathematical model for predicting the deterioration amount and learn a machine learning model for predicting the deterioration amount. The first usage history acquisition unit 101, the first deterioration prediction unit 102, the first feature amount calculation unit 103, the second deterioration prediction unit 104, the output unit 105, the mathematical model storage unit 201, and the machine learning model storage unit 202 perform the prediction process. The second usage history acquisition unit 111, the first mathematical model creation unit 112, the third deterioration prediction unit 113, the second feature amount calculation unit 114, the learning unit 115, the mathematical model storage unit 201, and the machine learning model storage unit 202 perform the learning process.

[0053] The first usage history acquisition unit 101 acquires the first usage history of the first battery 11. The communication unit (not shown) receives the first usage history transmitted by the first battery 11. The first usage history acquisition unit 101 acquires the first usage history received by the communication unit. The first usage history is the usage history of the first battery 11 from the initial state to the present. The initial state is the time when use began, and the present state is the time of the most recent measurement. Note that the terminal 30 may accept input of the first usage history by the user, and the communication unit (not shown) may receive the first usage history input by the terminal 30. Note that a memory (not shown) may store the first usage history, and the first usage history acquisition unit 101 may acquire the first usage history from the memory.

[0054] The mathematical model storage unit 201 stores a first mathematical model constructed from a theoretical formula derived from the chemical degradation characteristics of the battery.

[0055] The first deterioration prediction unit 102 acquires a first predicted deterioration amount of the first battery 11 by inputting the first usage history into a first mathematical model. The first deterioration prediction unit 102 calculates a first predicted deterioration amount of the first battery 11 by inputting the first usage history acquired by the first usage history acquisition unit 101 into a first mathematical model stored in the mathematical model storage unit 201. The first predicted deterioration amount represents a deterioration change amount ΔSOH of the first battery 11 from an initial state to the present. The deterioration change amount ΔSOH is the difference between the deterioration amount SOH (e.g., 100%) at the initial state (at the start of use) and the current deterioration amount SOH. For example, if the initial deterioration amount SOH is 100% and the current deterioration amount SOH is 90%, the first predicted deterioration amount (deterioration change amount) ΔSOH is 10%. Here, the first mathematical model for predicting the deterioration change amount (first predicted deterioration amount) ΔSOH will be described.

[0056] The first mathematical model is expressed by the following equation (1): The deterioration change amount (first predicted deterioration amount) ΔSOH is a value obtained by subtracting the deterioration amount SOH from 100%.

[0057] ΔSOH=ΔSOH Ａ +ΔSOH Ｂ ...(1)

[0058] In the above equation (1), ΔSOH represents the predicted deterioration amount of the battery, that is, the ratio of the current full charge capacity to the initial full charge capacity when the battery is first used, and ΔSOH Ａ represents the predicted storage deterioration amount due to storage deterioration, and ΔSOH Ｂ represents the predicted cycle deterioration amount due to cycle deterioration. The predicted deterioration amount ΔSOH of the battery is the predicted storage deterioration amount ΔSOH Ａ and the predicted cycle deterioration amount ΔSOH Ｂ The predicted deterioration amount ΔSOH and the predicted storage deterioration amount ΔSOH Ａ , and predicted cycle deterioration amount ΔSOH Ｂ represents the amount of change from the initial state.

[0059] Predicted storage deterioration amount ΔSOH Ａ is expressed by the following equation (2), and the predicted cycle deterioration amount ΔSOH Ｂ is expressed by the following equation (3).

[0060] ΔSOHＡ =A*exp(-E / RT)*t １／２ *exp(a・soc)...(2)

[0061] ΔSOH Ｂ = B * C ＤＯＤ／１００ ...(3)

[0062] In equation (2), A, E, and a are constants (or parameters), R is the gas constant, T is the absolute temperature, t is the elapsed time, and soc is the charge rate. In equation (3), B and C are constants (or parameters), and DOD is the depth of discharge. DOD is expressed as 0-100%, and represents the percentage of discharge from a fully charged state.

[0063] The first mathematical model includes a theoretical formula for calculating a predicted deterioration amount due to storage deterioration and a theoretical formula for calculating a predicted deterioration amount due to cycle deterioration.

[0064] In this embodiment, a memory (not shown) may store a first predicted deterioration amount of the first battery 11 obtained by inputting the first usage history into the first mathematical model. The first deterioration prediction unit 102 may acquire, from the memory, the first predicted deterioration amount of the first battery 11 obtained by inputting the first usage history into the first mathematical model.

[0065] The machine learning model storage unit 202 stores trained machine learning models. The machine learning models are trained based on feature quantities calculated from the battery's usage history, output values ​​of the mathematical model, and ground truth values ​​(measured values). The machine learning models are trained using the feature quantities calculated from the battery's usage history and the predicted battery deterioration amount calculated by inputting the usage history into a first mathematical model as explanatory variables, and the actually measured battery deterioration amount as a dependent variable. When the machine learning model receives the first feature quantities calculated from the first usage history and the first predicted deterioration amount of the first battery 11 calculated by inputting the first usage history into the first mathematical model, it outputs a second predicted deterioration amount of the first battery 11. The learning of the machine learning model will be described later.

[0066] The first feature calculation unit 103 calculates a first feature from the first usage history, and the first feature includes a cumulative elapsed time of the first battery 11 used under each of a plurality of usage conditions represented by a combination of a predetermined range of SOC, a predetermined range of current rate, and a predetermined range of temperature of the first battery 11.

[0067] For example, the SOC is divided into a range between 0% and 100% in increments of 5%. The current rate is divided into a range between 0 and less than -1, a range between -1 and 1 in increments of 0.1, and a range of 1 or greater. The temperature is divided into a range of 0 degrees or less, a range between 0°C and 60°C in increments of 5°C, and a range of 60 degrees or greater. Note that the above-mentioned predetermined ranges of SOC, current rate, and temperature are merely examples.

[0068] For example, based on the first usage history, the first characteristic calculation unit 103 calculates as the first characteristic the cumulative elapsed time of the first battery 11 used under usage conditions represented by a combination of an SOC in the range of 90-95%, a current rate in the range of 0.1-0.2, and a temperature in the range of 20-25 degrees.

[0069] The current rate is divided into 0 and a predetermined range other than 0. The causes of deterioration are significantly different between storage deterioration, which refers to deterioration when the battery is not charged or discharged, and cycle deterioration, which refers to deterioration when the battery is charged or discharged. In storage deterioration, the potential difference between the electrodes caused by the SOC during storage causes a reaction between the negative electrode (graphite) and the electrolyte, resulting in the formation of a solid electrolyte interphase (SEI) film (lithium compound), and the reduction in lithium ions leads to capacity deterioration. On the other hand, in cycle deterioration, capacity deterioration is primarily caused by the expansion and contraction of the positive electrode due to charging and discharging, resulting in breakdown. Other causes of capacity deterioration include lithium metal precipitation or electrolyte deterioration. To prevent the differences between the storage deterioration and cycle deterioration characteristics of a battery from being studied together, the current rate is divided into 0 and a predetermined range other than 0.

[0070] The second deterioration prediction unit 104 outputs a second predicted deterioration amount of the first battery 11 by inputting the first feature amount calculated from the first usage history and the first predicted deterioration amount into a machine learning model. The second deterioration prediction unit 104 calculates a second predicted deterioration amount ΔSOH of the first battery 11 by inputting the first feature amount calculated from the first usage history by the first feature amount calculation unit 103 and the first predicted deterioration amount calculated by the first deterioration prediction unit 102 into the machine learning model. The second deterioration prediction unit 104 inputs the first feature amount calculated from the first usage history and the first predicted deterioration amount, which is an output value from the first mathematical model, into the machine learning model and obtains a second predicted deterioration amount output from the machine learning model. The second predicted deterioration amount represents the deterioration change amount ΔSOH from the initial state to the present.

[0071] The output unit 105 outputs the second predicted deterioration amount calculated by the second deterioration prediction unit 104. The output unit 105 transmits the second predicted deterioration amount to the terminal 30 via a communication unit (not shown). The terminal 30 receives the second predicted deterioration amount of the first battery 11 transmitted by the information processing device 20 and displays the received second predicted deterioration amount of the first battery 11. The second predicted deterioration amount ΔSOH is a value obtained by subtracting the current predicted deterioration amount SOH from the initial deterioration amount SOH (e.g., 100%). Therefore, the output unit 105 may calculate the current predicted deterioration amount SOH by subtracting the second predicted deterioration amount ΔSOH from the initial deterioration amount SOH. Then, the output unit 105 may output the calculated current predicted deterioration amount SOH.

[0072] The second usage history acquisition unit 111 acquires the second usage history of the second battery 12, which is different from the first battery 11. A communication unit (not shown) receives the second usage history transmitted by the second battery 12. The second usage history acquisition unit 111 acquires the second usage history received by the communication unit. The second usage history is the usage history of the second battery 12 from the initial state to the present. The initial state is the time when use began, and the present state is the time of the most recent measurement. Note that the terminal 30 may accept input of the second usage history by the user, and the communication unit (not shown) may receive the second usage history input by the terminal 30. Note that a memory (not shown) may store the second usage history, and the second usage history acquisition unit 111 may acquire the second usage history from the memory.

[0073] Fig. 2 is a diagram showing an example of a usage history in the present embodiment 1. The usage history shown in Fig. 2 has contents common to the first usage history and the second usage history.

[0074] The usage history includes the date and time, current, voltage, capacity, temperature, SOC, and measured SOH. The measured SOH is the SOH that is actually measured. The interval at which the SOH is measured is longer than the interval at which the current, voltage, capacity, temperature, and SOC are measured. Therefore, the usage history for the period until the next SOH measurement includes the same measured SOH value. The initial date and time t (when the battery is first used) ０ The initial value of the measured SOH at ０ and soh ０ is 100%. １ ~Date and time t ３ The measured SOH at １ At date and time t ４ and date and time t ６ At time t, the SOH is measured. ４ ~Date and time t ５ The measured SOH at ２ and at time t ６ The measured SOH at ３ is.

[0075] The measured ΔSOH is a value obtained by subtracting the measured SOH from the initial measured SOH (for example, 100%). １ ~Date and time t ３ The measured ΔSOH at ｉ (soh ０ -soh １ ) and at time t ４ ~Date and time t ５ The measured ΔSOH at ｉｉ (soh ０ -soh ２ ) and at time t ６ The measured ΔSOH at ｉｉｉ (soh ０ -soh ３ The measured ΔSOH may be included in the usage history, or may be calculated based on the measured SOH acquired by the second usage history acquisition unit 111.

[0076] The first mathematical model creation unit 112 creates a first mathematical model based on the second usage history acquired by the second usage history acquisition unit 111. The first mathematical model creation unit 112 calculates parameters of the first mathematical model based on the second usage history acquired by the second usage history acquisition unit 111. The first mathematical model is expressed by the above equation (1). The parameters are the constants (A, E, a, B, and C) in the above equations (2) and (3).

[0077] Predicted storage deterioration amount ΔSOH due to storage deterioration Ａ is the predicted storage deterioration amount ΔSOH between each record of the second usage history Ａ (t ｘ＋１ -t ｘ , T ｘ , soc ｘ ) is calculated by adding all of the predicted cycle deterioration amounts ΔSOH Ｂ is the predicted cycle deterioration amount ΔSOH for each charge / discharge cycle in a predetermined period Ｂ It is calculated by integrating the above.

[0078] The first mathematical model creation unit 112 creates an equation for calculating the predicted deterioration amount ΔSOH for each SOH measured periodically, and calculates parameters using the least squares method that minimize the sum of squares of the difference between the predicted value of ΔSOH and the measured value.

[0079] The first mathematical model creation unit 112 stores the created first mathematical model in the mathematical model storage unit 201 .

[0080] The third deterioration prediction unit 113 calculates a third predicted deterioration amount of the second battery 12 by inputting the second usage history acquired by the second usage history acquisition unit 111 into the first mathematical model created by the first mathematical model creation unit 112. Every time the deterioration amount SOH of the second usage history is measured, the third deterioration prediction unit 113 calculates a third predicted deterioration amount ΔSOH of the second battery 12 using the second usage history for the period from the initial stage until the deterioration amount SOH of the second battery 12 is measured. The third deterioration prediction unit 113 calculates the third predicted deterioration amount ΔSOH for each period from the initial stage until the deterioration amount SOH of the second battery 12 is measured. In FIG. 2 , the predicted ΔSOH is a value calculated using the first mathematical model. At date and time t １ The predicted ΔSOH in Ｐ１ and at time t ４ The predicted ΔSOH in Ｐ２ and at time t ６ The predicted ΔSOH in Ｐ３ is.

[0081] The second feature amount calculation unit 114 calculates the second feature amount from the second usage history acquired by the second usage history acquisition unit 111. The second feature amount includes the cumulative elapsed time of the second battery 12 used under each of a plurality of usage conditions represented by combinations of a predetermined range of SOC, a predetermined range of current rate, and a predetermined range of temperature of the second battery 12. The second feature amount calculation unit 114 calculates the second feature amount for each predetermined period. The predetermined period is the period from the initial time to the date and time when the SOH is measured. Note that the predetermined period may be the measurement interval of the SOH.

[0082] The learning unit 115 uses the second feature amount calculated from the second usage history by the second feature amount calculation unit 114 and the third predicted deterioration amount calculated by the third deterioration prediction unit 113 as explanatory variables, and the actually measured deterioration amount of the second battery 12 as a dependent variable, to learn the machine learning model stored in the machine learning model storage unit 202. The learning unit 115 performs learning using, for example, a Light Gradient Boosting Machine (LightGBM).

[0083] FIG. 3 is a diagram showing an example of explanatory variables and response variables in the first embodiment.

[0084] The explanatory variables are the second feature amount calculated from the second usage history and the third predicted deterioration amount (predicted ΔSOH) which is a calculated value of the first mathematical model, and the dependent variable is the actually measured deterioration amount (measured ΔSOH) of the second battery 12. The second feature amount is calculated for each period from the initial time to the date and time when the SOH is measured for the Nth time. For example, in FIG. 2, the period P1 (t １ ~t ３ ), P2(t １ ~t ５ ), and P3(t １ ~t ６ ) a second feature amount is calculated.

[0085] In addition, the second characteristic value may be calculated not only for each period from the initial period to the date and time when the SOH is measured for the Nth time, but also for each period from the date and time when the SOH is measured for the Nth time to the date and time when the SOH is measured for the N+1th time.

[0086] In the example of FIG. 3, the period P1 (t １ ~t ３ ), the cumulative elapsed time at SOC 90-95%, current rate 0.1-0.2C, and temperature 20-25°C is 13.5 hours, the cumulative elapsed time at SOC 50-55%, current rate 0C, and temperature 0-5°C is 21 hours, the third predicted deterioration amount (predicted ΔSOH) is 0.8, and the measured deterioration amount (measured ΔSOH) is 0.5.

[0087] The learning unit 115 learns a machine learning model using the second feature amount for each predetermined period and the third predicted deterioration amount for each predetermined period as explanatory variables, and the measured deterioration amount of the second battery 12 for each predetermined period as a response variable. For example, the learning unit 115 learns a machine learning model using the second feature amount for each predetermined period and the third predicted deterioration amount for each predetermined period as explanatory variables. １ ~t ３ ) and the second feature amount of the period P1(t １ ~t ３ ) and the third predicted deterioration amount during the period P1(t １ ~t ３ The machine learning model is trained using the measured deterioration amount of the second battery 12 as the dependent variable.

[0088] The learning unit 115 stores the learned machine learning model in the machine learning model storage unit 202.

[0089] Next, the prediction process performed by the information processing device 20 in the first embodiment will be described.

[0090] FIG. 4 is a flowchart for explaining the prediction process performed by the information processing device 20 according to the first embodiment.

[0091] First, in step S1 , the first usage history acquisition unit 101 acquires the first usage history of the first battery 11 .

[0092] Next, in step S2, the first deterioration prediction unit 102 calculates the first predicted deterioration amount of the first battery 11 by inputting the first usage history acquired by the first usage history acquisition unit 101 into the first mathematical model stored in the mathematical model memory unit 201.

[0093] Next, in step S3, the first feature amount calculation unit 103 calculates a first feature amount based on the first usage history acquired by the first usage history acquisition unit 101. Based on the first usage history from the beginning to the present (the latest measurement point) acquired by the first usage history acquisition unit 101, the first feature amount calculation unit 103 calculates a first feature amount of the first battery 11, including the cumulative elapsed time for each predetermined period of time of the first battery 11 used under each of a plurality of usage conditions represented by combinations of a predetermined range of SOC, a predetermined range of current rate, and a predetermined range of temperature.

[0094] Next, in step S4, the second deterioration prediction unit 104 calculates the second predicted deterioration amount of the first battery 11 by inputting the first feature calculated from the first usage history by the first feature calculation unit 103 and the first predicted deterioration amount calculated by the first deterioration prediction unit 102 into a machine learning model.

[0095] Next, in step S5 , the output unit 105 outputs the second predicted deterioration amount calculated by the second deterioration prediction unit 104 .

[0096] In this way, a first predicted deterioration amount is obtained using a first mathematical model constructed using a theoretical formula derived from the chemical deterioration characteristics of the battery, and the obtained first predicted deterioration amount is input into a machine learning model to calculate a second predicted deterioration amount of the first battery 11, thereby improving the accuracy of predicting deterioration of a battery used under conditions outside the range learned by the machine learning model.

[0097] Next, the learning process performed by the information processing device 20 in the first embodiment will be described.

[0098] FIG. 5 is a flowchart for explaining the learning process performed by the information processing device 20 according to the first embodiment.

[0099] First, in step S11 , the second usage history acquisition unit 111 acquires the second usage history of the second battery 12 that is different from the first battery 11 .

[0100] Next, in step S12 , the first mathematical model creation unit 112 creates a first mathematical model based on the second usage history acquired by the second usage history acquisition unit 111 .

[0101] Next, in step S13 , the first mathematical model creation unit 112 stores the created first mathematical model in the mathematical model storage unit 201 .

[0102] Next, in step S14, the third deterioration prediction unit 113 calculates the third predicted deterioration amount of the second battery 12 by inputting the second usage history acquired by the second usage history acquisition unit 111 into the first mathematical model created by the first mathematical model creation unit 112.

[0103] Next, in step S15, the second feature amount calculation unit 114 calculates a second feature amount based on the second usage history acquired by the second usage history acquisition unit 111. Based on the second usage history from the beginning to the present (the latest measurement point) acquired by the second usage history acquisition unit 111, the second feature amount calculation unit 114 calculates a second feature amount of the second battery 12, including the cumulative elapsed time for each predetermined period of time of the second battery 12 used under each of a plurality of usage conditions represented by a combination of a predetermined range of SOC, a predetermined range of current rate, and a predetermined range of temperature.

[0104] Next, in step S16, the learning unit 115 learns the machine learning model stored in the machine learning model memory unit 202 using the second feature calculated from the second usage history by the second feature calculation unit 114 and the third predicted deterioration amount calculated by the third deterioration prediction unit 113 as explanatory variables, and the measured deterioration amount of the second battery 12 actually measured and included in the second usage history as the objective variable.

[0105] Next, in step S17 , the learning unit 115 stores the learned machine learning model in the machine learning model storage unit 202 .

[0106] (Embodiment 2) In the learning process of embodiment 1, a first mathematical model is created and a machine learning model is trained using the second usage history of second battery 12. In contrast, in the learning process of embodiment 2, a first mathematical model is created and a machine learning model is trained using the second usage history of second battery 12 and a third usage history of third battery 13, which is different from first battery 11 and second battery 12.

[0107] FIG. 6 is a diagram showing the configuration of an information processing system 1A according to the second embodiment.

[0108] The information processing system 1A includes a first battery 11, a second battery 12, a third battery 13, an information processing device 20A, and a terminal 30. In the information processing system 1A according to the second embodiment, the same components as those in the information processing system 1 according to the first embodiment are denoted by the same reference numerals, and descriptions thereof will be omitted.

[0109] The information processing system 1A includes a first battery 11, a second battery 12, a third battery 13, an information processing device 20A, and a terminal 30. In the information processing system 1A according to the second embodiment, the same components as those in the information processing system 1 according to the first embodiment are denoted by the same reference numerals, and descriptions thereof will be omitted.

[0110] The third battery 13 is a battery different from the first battery 11 and the second battery 12. The third battery 13 is a secondary battery used for creating the first mathematical model and training the machine learning model, and stores power by charging and supplies power by discharging. The third battery 13 is connected to the information processing device 20A via a network so that they can communicate with each other.

[0111] The first battery 11, the second battery 12, and the third battery 13 may be batteries manufactured by different manufacturers. The first battery 11, the second battery 12, and the third battery 13 are preferably the same type of secondary battery (e.g., lithium ion battery). The first battery 11, the second battery 12, and the third battery 13 may have different capacities.

[0112] The first battery 11, the second battery 12, and the third battery 13 may be different types of lithium ion batteries, such as a cobalt-based lithium ion battery, a manganese-based lithium ion battery, a nickel-based lithium ion battery, an NCA-based lithium ion battery, an iron phosphate-based lithium ion battery, or a ternary lithium ion battery.

[0113] Furthermore, the secondary battery is not limited to a lithium ion battery, but may be a lead storage battery, a nickel-metal hydride battery, an all-solid-state battery, or the like.

[0114] The third battery 13 transmits a third usage history to the information processing device 20A. The third usage history includes the date and time, current value, voltage value, capacity, temperature, SOC, and SOH. The third battery 13 periodically measures the current value, voltage value, capacity, temperature, SOC, and SOH. The content of the third usage history is the same as the content of the first usage history.

[0115] Furthermore, the second battery 12 and the third battery 13 are used in different ways.

[0116] The information processing device 20A includes a first usage history acquisition unit 101, a first deterioration prediction unit 102, a first feature calculation unit 103, a second deterioration prediction unit 104, an output unit 105, a second usage history acquisition unit 111, a second feature calculation unit 114, a learning unit 115A, a second mathematical model creation unit 116, a third usage history acquisition unit 117, a third mathematical model creation unit 118, a fourth deterioration prediction unit 119, a fifth deterioration prediction unit 120, a third feature calculation unit 121, a fourth mathematical model creation unit 122, a mathematical model memory unit 201, and a machine learning model memory unit 202.

[0117] The information processing device 20A includes at least a computer system including, for example, a control program, a processing circuit such as a processor or logic circuit that executes the control program, and a recording device such as an internal memory or an accessible external memory that stores the control program. Note that the information processing device 20A may be realized, for example, by hardware implementation using the processing circuit, or by execution of a software program stored in the memory by the processing circuit or distributed from an external server, or by a combination of these hardware and software implementations.

[0118] The information processing device 20A may be a server or an edge computer. The configuration of the information processing device 20A may be implemented in the terminal 30. The configuration of the information processing device 20A may be distributed among the terminal 30 and the server.

[0119] The information processing device 20A performs a prediction process to predict the deterioration amount of the first battery 11 and a learning process to create a first mathematical model for predicting the deterioration amount and learn a machine learning model for predicting the deterioration amount. The first usage history acquisition unit 101, the first deterioration prediction unit 102, the first feature amount calculation unit 103, the second deterioration prediction unit 104, the output unit 105, the mathematical model storage unit 201, and the machine learning model storage unit 202 perform the prediction process. The second usage history acquisition unit 111, the second feature amount calculation unit 114, the learning unit 115A, the second mathematical model creation unit 116, the third usage history acquisition unit 117, the third mathematical model creation unit 118, the fourth deterioration prediction unit 119, the fifth deterioration prediction unit 120, the third feature amount calculation unit 121, the fourth mathematical model creation unit 122, the mathematical model storage unit 201, and the machine learning model storage unit 202 perform the learning process.

[0120] The second mathematical model creation unit 116 creates a second mathematical model based on the second usage history acquired by the second usage history acquisition unit 111. The second mathematical model creation unit 116 calculates parameters of the second mathematical model based on the second usage history acquired by the second usage history acquisition unit 111. The second mathematical model is expressed by the above equation (1). The parameters are the constants (A, E, a, B, and C) in the above equations (2) and (3). The method for creating the second mathematical model is the same as the method for creating the first mathematical model.

[0121] The third usage history acquisition unit 117 acquires a third usage history of the third battery 13, which is different from the first battery 11 and the second battery 12. A communication unit (not shown) receives the third usage history transmitted by the third battery 13. The third usage history acquisition unit 117 acquires the third usage history received by the communication unit. The third usage history is a usage history of the third battery 13 from an initial time to the present. The initial time is the time when use began, and the present time is the time of the most recent measurement. Note that the terminal 30 may accept input of the third usage history by the user, and the communication unit (not shown) may receive the third usage history input by the terminal 30. Note that a memory (not shown) may store the third usage history, and the third usage history acquisition unit 117 may acquire the third usage history from the memory.

[0122] The third mathematical model creation unit 118 creates a third mathematical model based on the third usage history acquired by the third usage history acquisition unit 117. The third mathematical model creation unit 118 calculates parameters of the third mathematical model based on the third usage history acquired by the third usage history acquisition unit 117. The third mathematical model is expressed by the above equation (1). The parameters are the constants (A, E, a, B, and C) in the above equations (2) and (3). The method for creating the third mathematical model is the same as the method for creating the first mathematical model.

[0123] The fourth deterioration prediction unit 119 calculates a fourth predicted deterioration amount of the third battery 13 by inputting the third usage history acquired by the third usage history acquisition unit 117 into the second mathematical model created by the second mathematical model creation unit 116. Each time the deterioration amount SOH of the third usage history is measured, the fourth deterioration prediction unit 119 calculates a fourth predicted deterioration amount ΔSOH of the third battery 13 using the second usage history for the period from the initial stage until the deterioration amount SOH of the third battery 13 is measured. The fourth deterioration prediction unit 119 calculates the fourth predicted deterioration amount ΔSOH for each period from the initial stage until the deterioration amount SOH of the third battery 13 is measured.

[0124] The fifth deterioration prediction unit 120 calculates a fifth predicted deterioration amount of the second battery 12 by inputting the second usage history acquired by the second usage history acquisition unit 111 into the third mathematical model created by the third mathematical model creation unit 118. Each time the deterioration amount SOH of the second usage history is measured, the fifth deterioration prediction unit 120 calculates a fifth predicted deterioration amount ΔSOH of the second battery 12 using the second usage history for the period from the initial stage until the deterioration amount SOH of the second battery 12 is measured. The fifth deterioration prediction unit 120 calculates the fifth predicted deterioration amount ΔSOH for each period from the initial stage until the deterioration amount SOH of the third battery 13 is measured.

[0125] The third feature amount calculation unit 121 calculates the third feature amount from the third usage history acquired by the third usage history acquisition unit 117. The third feature amount includes the cumulative elapsed time of the third battery 13 used under each of a plurality of usage conditions represented by combinations of a predetermined range of SOC, a predetermined range of current rate, and a predetermined range of temperature of the third battery 13. The third feature amount calculation unit 121 calculates the third feature amount for each predetermined period. The predetermined period is the period from the initial time to the date and time when the SOH is measured. Note that the predetermined period may be the measurement interval of the SOH.

[0126] The learning unit 115A uses the third feature amount calculated from the third usage history by the third feature amount calculation unit 121 and the fourth predicted deterioration amount calculated by the fourth deterioration prediction unit 119 as explanatory variables, and the actually measured deterioration amount of the third battery 13 as a dependent variable, to learn the machine learning model stored in the machine learning model storage unit 202. The learning unit 115A performs learning using, for example, LightGBM. The learning unit 115A uses the third feature amount for each predetermined period and the fourth predicted deterioration amount for each predetermined period as explanatory variables, and learns the machine learning model using the measured deterioration amount of the third battery 13 for each predetermined period as a dependent variable.

[0127] Furthermore, the learning unit 115A uses the second feature amount calculated from the second usage history by the second feature amount calculation unit 114 and the fifth predicted deterioration amount calculated by the fifth deterioration prediction unit 120 as explanatory variables, and the actually measured deterioration amount of the second battery 12 as a dependent variable, to learn the machine learning model stored in the machine learning model storage unit 202. The learning unit 115A performs learning using, for example, LightGBM. The learning unit 115A uses the second feature amount for each predetermined period and the fifth predicted deterioration amount for each predetermined period as explanatory variables, and learns the machine learning model using the measured deterioration amount of the second battery 12 for each predetermined period as a dependent variable.

[0128] The fourth mathematical model creation unit 122 creates a first mathematical model based on the second usage history acquired by the second usage history acquisition unit 111 and the third usage history acquired by the third usage history acquisition unit 117. The fourth mathematical model creation unit 122 calculates parameters of the first mathematical model based on the second usage history acquired by the second usage history acquisition unit 111 and the third usage history acquired by the third usage history acquisition unit 117. The first mathematical model is expressed by the above equation (1). The parameters are the constants (A, E, a, B, and C) in the above equations (2) and (3). Note that the method of creating the first mathematical model in the second embodiment is the same as the method of creating the first mathematical model in the first embodiment.

[0129] The fourth mathematical model creation unit 122 stores the created first mathematical model in the mathematical model storage unit 201 .

[0130] The prediction process by the information processing device 20A in the second embodiment is the same as the prediction process by the information processing device 20 in the first embodiment.

[0131] Next, the learning process performed by the information processing device 20A in the second embodiment will be described.

[0132] FIG. 7 is a flowchart for explaining the learning process by the information processing device 20A according to the second embodiment.

[0133] First, in step S21, the second usage history acquisition unit 111 acquires the second usage history of the second battery 12 that is different from the first battery 11.

[0134] Next, in step S22 , the third usage history acquisition unit 117 acquires a third usage history of the third battery 13 , which is different from the first battery 11 and the second battery 12 .

[0135] Next, in step S23 , the second mathematical model creation unit 116 creates a second mathematical model based on the second usage history acquired by the second usage history acquisition unit 111 .

[0136] Next, in step S24, the fourth deterioration prediction unit 119 calculates the fourth predicted deterioration amount of the third battery 13 by inputting the third usage history acquired by the third usage history acquisition unit 117 into the second mathematical model created by the second mathematical model creation unit 116.

[0137] Next, in step S25, the third characteristic amount calculation unit 121 calculates a third characteristic amount based on the third usage history acquired by the third usage history acquisition unit 117. Based on the third usage history from the beginning to the present (the latest measurement point) acquired by the third usage history acquisition unit 117, the third characteristic amount calculation unit 121 calculates a third characteristic amount of the third battery 13, including the cumulative elapsed time for each predetermined period of time of the third battery 13 used under each of a plurality of usage conditions represented by combinations of a predetermined range of SOC, a predetermined range of current rate, and a predetermined range of temperature.

[0138] Next, in step S26, the learning unit 115A learns the machine learning model stored in the machine learning model memory unit 202 using the third feature calculated from the third usage history by the third feature calculation unit 121 and the fourth predicted deterioration amount calculated by the fourth deterioration prediction unit 119 as explanatory variables, and the measured deterioration amount of the third battery 13 actually measured and included in the third usage history as the objective variable.

[0139] Next, in step S27 , the third mathematical model creation unit 118 creates a third mathematical model based on the third usage history acquired by the third usage history acquisition unit 117 .

[0140] Next, in step S28, the fifth deterioration prediction unit 120 calculates the fifth predicted deterioration amount of the second battery 12 by inputting the second usage history acquired by the second usage history acquisition unit 111 into the third mathematical model created by the third mathematical model creation unit 118.

[0141] Next, in step S29, the second feature amount calculation unit 114 calculates a second feature amount based on the second usage history acquired by the second usage history acquisition unit 111. Based on the second usage history from the beginning to the present (the latest measurement point) acquired by the second usage history acquisition unit 111, the second feature amount calculation unit 114 calculates a second feature amount of the second battery 12, including the cumulative elapsed time for each predetermined period of time of the second battery 12 used under each of a plurality of usage conditions represented by a combination of a predetermined range of SOC, a predetermined range of current rate, and a predetermined range of temperature.

[0142] Next, in step S30, the learning unit 115A learns the machine learning model stored in the machine learning model memory unit 202 using the second feature calculated from the second usage history by the second feature calculation unit 114 and the fifth predicted deterioration amount calculated by the fifth deterioration prediction unit 120 as explanatory variables, and the actually measured deterioration amount of the second battery 12 as the objective variable.

[0143] Next, in step S31, the learning unit 115A stores the learned machine learning model in the machine learning model storage unit 202.

[0144] Next, in step S32, the fourth mathematical model creation unit 122 creates a first mathematical model based on the second usage history acquired by the second usage history acquisition unit 111 and the third usage history acquired by the third usage history acquisition unit 117.

[0145] Next, in step S33 , the fourth mathematical model creation unit 122 stores the created first mathematical model in the mathematical model storage unit 201 .

[0146] In this second embodiment, the first mathematical model is created and the machine learning model is trained using the usage histories of two batteries, the second battery 12 and the third battery 13, but the present disclosure is not limited to this, and the first mathematical model may be created and the machine learning model may be trained using the usage histories of three or more batteries.

[0147] Furthermore, in the first and second embodiments, the first predicted deterioration amount and the second predicted deterioration amount may be calculated for the period from the initial stage to the present, or the first predicted deterioration amount and the second predicted deterioration amount may be calculated for the period from the initial stage to a predetermined prediction point in the future.

[0148] In this case, the first usage history acquisition unit 101 may create an estimated first usage history of the first battery 11 from the present (latest measurement point) to a predetermined prediction point in the future based on the acquired first usage history. For example, when predicting the deterioration amount of the first battery 11 one month from the present, the first usage history acquisition unit 101 copies the first usage history for the past one month from the present (latest measurement point) and changes the date and time of the copied one month of first usage history to a date and time from the present (latest measurement point) to one month from now. In this way, the first usage history acquisition unit 101 creates an estimated usage history from the present (latest measurement point) to one month from now. The first usage history acquisition unit 101 may add the created estimated usage history to the acquired first usage history to create an estimated first usage history from the beginning (the point at which use begins) to the prediction point.

[0149] The first deterioration prediction unit 102 may input the estimated first usage history into a first mathematical model to calculate a first predicted amount of deterioration of the first battery 11 during a period from an initial stage to a predetermined prediction point in time in the future. The second deterioration prediction unit 104 may input the estimated first feature amount calculated from the estimated first usage history and the first predicted amount of deterioration into a machine learning model to calculate a second predicted amount of deterioration of the first battery 11 during a period from an initial stage to a predetermined prediction point in time in the future.

[0150] Note that some or all of the functions of the device according to the embodiment of the present disclosure may be realized by a processor such as a CPU executing a program.

[0151] Furthermore, all the numbers used above are merely examples to specifically explain the present disclosure, and the present disclosure is not limited to the numbers used as examples.

[0152] The order in which the steps are executed in the above flowchart is merely an example for specifically explaining the present disclosure, and other orders may be used as long as similar effects are obtained. Also, some of the steps may be executed simultaneously (in parallel) with other steps.

[0153] The technology disclosed herein is useful as a technology for predicting the amount of battery degradation because it can improve the accuracy of predicting the degradation of batteries used under conditions outside the range learned by the machine learning model.

Claims

1. An information processing method executed by a computer, comprising: acquiring a first usage history of a first battery; acquiring a first predicted deterioration amount of the first battery by inputting the first usage history into a first mathematical model; and outputting a second predicted deterioration amount of the first battery by inputting a first feature calculated from the first usage history and the first predicted deterioration amount into a machine learning model.

2. The information processing method according to claim 1, wherein the first mathematical model includes a theoretical formula for calculating a predicted amount of deterioration due to storage deterioration and a theoretical formula for calculating a predicted amount of deterioration due to cycle deterioration.

3. An information processing method according to claim 1 or 2, wherein the machine learning model is trained using, as explanatory variables, features calculated from the battery's usage history and the predicted amount of deterioration of the battery calculated by inputting the usage history into the first mathematical model, and the actually measured amount of deterioration of the battery as the objective variable.

4. An information processing method as described in claim 1 or 2, further comprising: acquiring a second usage history of a second battery different from the first battery; and creating the first mathematical model based on the second usage history.

5. The information processing method of claim 4, further comprising: calculating a third predicted deterioration amount of the second battery by inputting the second usage history into the first mathematical model; and training the machine learning model using the second feature calculated from the second usage history and the third predicted deterioration amount as explanatory variables and the actually measured deterioration amount of the second battery as a dependent variable.

6. The information processing method of claim 1 or 2, further comprising: acquiring a second usage history of a second battery different from the first battery; acquiring a third usage history of a third battery different from the first battery and the second battery; creating a second mathematical model based on the second usage history; calculating a fourth predicted deterioration amount of the third battery by inputting the third usage history into the second mathematical model; and training the machine learning model using a third feature calculated from the third usage history and the fourth predicted deterioration amount as explanatory variables and an actually measured deterioration amount of the third battery as a target variable.

7. The information processing method of claim 6, further comprising: creating a third mathematical model based on the third usage history; calculating a fifth predicted deterioration amount of the second battery by inputting the second usage history into the third mathematical model; and training the machine learning model using the second feature calculated from the second usage history and the fifth predicted deterioration amount as explanatory variables and the actually measured deterioration amount of the second battery as a dependent variable.

8. The information processing method according to claim 6, further comprising creating the first mathematical model based on the second usage history and the third usage history.

9. An information processing method according to claim 1 or 2, wherein the first characteristic amount includes a cumulative elapsed time of the first battery used under each of a plurality of usage conditions represented by a combination of a predetermined range of SOC, a predetermined range of current rate, and a predetermined range of temperature of the first battery.

10. An information processing device having a processor, wherein the processor: acquires a first usage history of a first battery; acquires a first predicted deterioration amount of the first battery by inputting the first usage history into a first mathematical model; and outputs a second predicted deterioration amount of the first battery by inputting a first feature calculated from the first usage history and the first predicted deterioration amount into a machine learning model.

11. An information processing program that causes a computer to function as follows: acquiring a first usage history of a first battery; acquiring a first predicted deterioration amount of the first battery by inputting the first usage history into a first mathematical model; and outputting a second predicted deterioration amount of the first battery by inputting a first feature calculated from the first usage history and the first predicted deterioration amount into a machine learning model.

Images