Prediction device, prediction method, information processing method, information processing device, and computer program

Abstract

The present invention provides a technology or the like capable of improving prediction accuracy of the state of a power storage element using a prediction model.　This prediction device comprises a processing unit that: acquires a first prediction value from a prediction model for predicting the state of a power storage element; by using a correction model for correcting an error in a prediction value from the prediction model, estimates a first correction value for correcting an error in the acquired first prediction value; and, by using the estimated first correction value, corrects the first prediction value.

Description

Prediction apparatus, prediction method, information processing method, information processing apparatus, and computer program

[0001] The present invention relates to a prediction apparatus, a prediction method, an information processing method, an information processing apparatus, and a computer program.

[0002] Electric storage elements are widely used in batteries for uninterruptible power supply (backup), batteries for mobile body power sources such as automobiles, railways, and aircraft, batteries for auxiliary machinery, batteries for renewable energy power plants that compensate for stable power supply, and the like.

[0003] Deterioration of an electric storage element progresses as it is repeatedly charged and discharged. In the operation of an electric storage element, it is important to accurately predict the deterioration state, service life, and the like of the electric storage element. For example, Patent Document 1 proposes a technique of deriving the service life of a secondary battery using a model that outputs a battery capacity when the usage status of the secondary battery is input.

[0004] Also, in the operation of an electric storage element, it is important to accurately grasp the service life of the electric storage element. Conventionally, techniques for predicting the service life of an electric storage element have been proposed (see, for example, Patent Document 2).

[0005] International Publication No. WO 2020 / 044713, Japanese Unexamined Patent Publication No. 2000-228227

[0006] In a prediction model constructed using data obtained from a test using a test electric storage element, an error may occur between the output of the prediction model and the actual state of the electric storage element due to factors such as the content of test data and the difference between the test electric storage element and an actual electric storage element, which may reduce prediction accuracy.

[0007] In one aspect, an object of the present invention is to provide a technique capable of improving the prediction accuracy of the state of an electric storage element using a prediction model.

[0008] In various models for predicting the state of an electric storage element, an error may occur between the output of the prediction model and the actual state of the electric storage element due to influences such as deterioration of the electric storage element and the usage environment. It is conceivable to correct prediction values obtained from various prediction models to improve prediction accuracy. However, sufficient studies have not yet been conducted on correction of prediction values obtained from various prediction models.

[0009] One aspect of this is the aim to provide information processing methods that can correct predicted values ​​predicted by various prediction models.

[0010] The technology described in Patent Document 2 has room for improvement in terms of the accuracy of predicting energy storage capacity.

[0011] One aspect of this project is to provide prediction methods that can improve the accuracy of predicting energy storage capacity.

[0012] A prediction device relating to one aspect of this disclosure includes a processing unit that acquires a first predicted value using a prediction model for predicting the state of an energy storage element, estimates a first correction value for correcting the error of the acquired first predicted value using a correction model for correcting the error of the prediction value of the prediction model, and corrects the first predicted value using the estimated first correction value.

[0013] An information processing method relating to one aspect of this disclosure involves an information processing device of the first business operator acquiring predicted values ​​predicted by a second prediction model of the second business operator for predicting the state of an energy storage element, generating correction information for the acquired predicted values ​​using a correction model constructed based on the error between the predicted values ​​predicted by the first prediction model of the first business operator for predicting the state of the energy storage element and the measured values ​​of the state of the energy storage element, and outputting the generated correction information.

[0014] A prediction method relating to one aspect of this disclosure involves obtaining the structure or application of an energy storage device, identifying a first prediction model from among a plurality of first prediction models for predicting conditions other than degradation in the energy storage device that corresponds to the type of structure or application obtained, and predicting the lifespan of the energy storage device based on the predicted state value of the energy storage device predicted by the identified first prediction model and the predicted degradation value of the energy storage device predicted by a second prediction model for predicting the degradation of the energy storage device.

[0015] From one perspective, it is possible to improve the accuracy of predicting the state of energy storage elements using predictive models.

[0016] From one perspective, it is possible to correct the predicted values ​​predicted by various prediction models.

[0017] From one perspective, this could improve the accuracy of predicting energy storage capacity.

[0018] This figure shows an overview of the prediction system of this embodiment. This block diagram shows an example of the configuration of the prediction device. This figure illustrates the method for predicting the state of energy storage elements performed by the prediction device. This flowchart shows an example of the processing procedure performed by the prediction device. This flowchart shows an example of the processing procedure performed by the prediction device of the second embodiment. This flowchart shows an example of the processing procedure performed by the prediction device of the second embodiment. This figure shows an overview of the energy storage information processing system of the third embodiment. This block diagram shows an example of the configuration of the information processing device. This figure illustrates a method for correcting predicted values ​​of the degradation state using a correction model. This flowchart shows an example of the procedure for reconstructing the correction model. This flowchart shows an example of the procedure for correcting predicted values ​​of the prediction model. This figure shows an overview of the prediction system of the fourth embodiment. This block diagram shows an example of the configuration of the prediction device. This figure shows an example of the contents of the information stored in the correspondence table. This flowchart shows an example of the processing procedure performed by the prediction device.

[0019] (1) A prediction device according to one aspect of the present disclosure includes a processing unit that acquires a first predicted value by a prediction model for predicting the state of an energy storage element, estimates a first correction value for correcting the error of the acquired first predicted value using a correction model for correcting the error of the prediction value of the prediction model, and corrects the first predicted value using the estimated first correction value.

[0020] Predictive models for predicting the state of energy storage elements (e.g., cells) typically determine their formulas and parameters based on test data obtained from durability tests using test cells. However, errors can occur between the predicted values ​​of the predictive model and the actual state of the cell due to various reasons such as insufficient test data, design differences between the actual cell and the test cell, changes in degradation characteristics during operation of the actual cell, and use under unexpected conditions. Eliminating these prediction errors is crucial to ensuring the accuracy of the predictive model.

[0021] Improving the prediction model itself can eliminate errors and improve prediction accuracy, but updating complex prediction models is not easy and is time-consuming. Optimizing a prediction model for specific cells reduces its versatility, making it difficult to apply to a variety of cells.

[0022] According to the prediction device described in (1) above, by preparing a correction model that corrects the predicted values ​​of the prediction model in addition to the prediction model itself, the predicted values ​​of the prediction model can be efficiently corrected using the correction values ​​from the correction model. Since the predicted values ​​can be corrected without changing the prediction model itself, the increase in processing load can be suppressed while improving prediction accuracy. Predicted values ​​tailored to individual real cells can be output with high accuracy while making use of the prediction characteristics of the pre-built prediction model.

[0023] (2) In the prediction device described in (1) above, the correction model may be a model that outputs a correction value according to the usage history data of the energy storage element.

[0024] According to the configuration described in (2) above, the correction model can accurately output correction values ​​to bring the predicted values ​​of the prediction model closer to the measured values, in accordance with the usage history of the energy storage element.

[0025] (3) The prediction device described in (1) or (2) above may update the prediction model if the error between the corrected first prediction value and the measured value of the state of the energy storage element is outside the acceptable range.

[0026] According to the configuration described in (3) above, if the error cannot be sufficiently eliminated by correction using the correction model, the prediction model itself is updated, thereby further improving prediction accuracy. By performing correction in advance using a correction model with low processing load, and then updating the prediction model with a higher processing load if the accuracy is insufficient, the state of the energy storage element can be predicted efficiently while ensuring prediction accuracy.

[0027] (4) If the prediction device described in (3) above updates the prediction model, it may update the correction model based on the predicted values ​​of the updated prediction model.

[0028] According to the configuration described in (4) above, the correction model can be updated based on the error between the new predicted values ​​and the measured values ​​that accompany the update of the prediction model. This makes it possible to construct a correction model that can accurately estimate the correction values ​​for the predicted values ​​of the updated prediction model.

[0029] (5) Any one of the prediction devices described in (1) to (4) above may determine whether the first error between the corrected first predicted value and the measured value of the state of the energy storage element is within an acceptable range, and if it determines that the first error is outside the acceptable range, it may update the parameters in the prediction model, which is shown by a calculation formula including predetermined parameters, and determine whether the second error between the second predicted value from the prediction model after updating the parameters and the measured value of the state of the energy storage element is within an acceptable range, and if it determines that the second error is outside the acceptable range, it may update the calculation formula in the prediction model.

[0030] According to the configuration described in (5) above, if the error cannot be sufficiently resolved by correction using the correction model, priority is given to updating parameters that have a low processing load among the updates to the prediction model, thereby efficiently improving prediction accuracy. By updating the calculation formula itself in addition to the parameters as needed, the accuracy of the prediction model can be further improved.

[0031] (6) If the prediction device described in (5) above determines that the second error is outside the acceptable range, it may estimate a second correction value for correcting the error in the second predicted value using the correction model, correct the second predicted value by the prediction model after updating the parameters using the estimated second correction value, determine whether the third error between the corrected second predicted value and the measured value of the state of the energy storage element is within the acceptable range, and if it determines that the third error is outside the acceptable range, it may update the calculation formula in the prediction model.

[0032] According to the configuration described in (6) above, the correction using the correction model and the update of the prediction model can be combined and processed in the appropriate order, thus enabling efficient and effective processing.

[0033] (7) Any one of the prediction devices described in (1) to (6) above may output information indicating an abnormal state of the energy storage element if the update frequency of the correction model or the prediction model, or the predicted value obtained using the updated correction model or the prediction model, matches the abnormal state.

[0034] If errors cannot be sufficiently resolved even after correcting the predicted values ​​or updating the model, there is a high probability that the energy storage element itself is faulty. According to the configuration in (7) above, such abnormal conditions can be detected and reported, making it possible to identify the abnormal condition early.

[0035] (8) A prediction method according to one aspect of the present disclosure involves obtaining a first predicted value using a prediction model that predicts the state of an energy storage element, estimating a first correction value for correcting the error of the obtained first predicted value using a correction model that corrects the error of the prediction value of the prediction model, and correcting the first predicted value using the estimated first correction value.

[0036] (9) A computer program according to one aspect of the present disclosure obtains a first predicted value using a prediction model for predicting the state of an energy storage element, estimates a first correction value for correcting the error of the obtained first predicted value using a correction model for correcting the error of the prediction value of the prediction model, and causes the computer to perform a process of correcting the first predicted value using the estimated first correction value.

[0037] (10) An information processing method according to one aspect of the present disclosure includes an information processing device of the first business operator that acquires predicted values ​​predicted by a second prediction model of the second business operator that predicts the state of an energy storage element, generates correction information for the acquired predicted values ​​using a correction model constructed based on the error between the predicted values ​​predicted by the first prediction model of the first business operator that predicts the state of an energy storage element and the measured value of the state of the energy storage element, and outputs the generated correction information.

[0038] According to the information processing method described in (10) above, the first business operator can use a correction model generated by error analysis of the first prediction model to correct the predicted values ​​from various prediction models held by other businesses. Conventionally, each business operator has been conducting research on prediction models, but no service has been considered to improve the prediction accuracy while utilizing each business operator's prediction model. According to the information processing method disclosed herein, the first business operator can use a pre-prepared correction model to provide a service for correcting predicted values ​​for prediction models from various businesses. By separately preparing a correction model to correct errors in prediction models, it is possible to efficiently obtain highly accurate predicted values ​​by utilizing the prediction characteristics of the prediction model without changing the prediction model itself.

[0039] (11) The information processing method in (10) above may involve obtaining the error between the predicted value predicted by the second prediction model and the measured value of the state of the energy storage element, reconstructing the correction model based on the obtained error between the predicted value of the second prediction model and the measured value, and generating the correction information using the reconstructed correction model.

[0040] According to the configuration described in (11) above, the correction model can be adjusted to match the prediction characteristics of the second prediction model that generates the predicted values, thereby improving the accuracy of the correction by the correction model.

[0041] (12) In the information processing method of (10) or (11) above, the correction model may be a model that takes operational data indicating the operational status of the energy storage element as input and outputs the error between the predicted value of a prediction model that predicts the state of the energy storage element and the measured value.

[0042] According to the configuration described in (12) above, a predicted value that is in line with the operating status of the energy storage element to be predicted can be obtained with high accuracy.

[0043] (13) The information processing method described in (11) or (12) above may obtain the predicted value predicted by the second prediction model provided by the second business operator and obtain the error provided by the first business operator.

[0044] According to the configuration of (13) above, even when error data necessary for reconstructing a correction model is not obtained at the second operator, the correction model can be reconstructed using the error data collected by the first operator, thereby reducing the data collection burden on the second operator and improving convenience.

[0045] (14) The information processing method according to any one of (10) to (13) above may comprise: acquiring design information of an electricity storage element used for constructing the second prediction model; acquiring an error between a predicted value predicted by the first prediction model and a measured value for a state of another electricity storage element having design information similar to the acquired design information; reconstructing the correction model based on the acquired error between the predicted value and the measured value of the first prediction model; and generating the correction information using the reconstructed correction model.

[0046] According to the configuration of (14) above, even when the second operator does not provide the predicted value of the second prediction model or error data necessary for reconstructing the correction model, the first operator can prepare data necessary for reconstructing the correction model and reconstruct the correction model by acquiring the design information. This can reduce the data collection burden on the second operator and improve convenience.

[0047] (15) The information processing method according to any one of (10) to (14) above may comprise: outputting the correction information for the predicted value acquired from the second operator to the second operator; and calculating a usage fee such that the larger the data amount of the predicted value acquired from the second operator is, the higher the usage fee becomes.

[0048] According to the configuration of (15) above, the usage fee can be varied according to the data amount of the prediction value that is a target for generating correction information, and the degree of freedom in service design is improved.

[0049] (16) A computer program according to one aspect of the present disclosure causes a computer to obtain predicted values ​​predicted by a second prediction model of a second business operator for predicting the state of an energy storage element, generate correction information for the obtained predicted values ​​using a correction model generated based on the error between the predicted values ​​predicted by a first prediction model of a first business operator for predicting the state of an energy storage element and the measured value of the state of the energy storage element, and output the generated correction information.

[0050] (17) An information processing device according to one aspect of the present disclosure includes a processing unit that acquires predicted values ​​predicted by a second prediction model of a second business operator for predicting the state of an energy storage element, generates correction information for the acquired predicted values ​​using a correction model generated based on the error between the predicted values ​​predicted by a first prediction model of a first business operator for predicting the state of an energy storage element and the measured value of the state of the energy storage element, and outputs the generated correction information.

[0051] (18) A prediction method according to one aspect of the present disclosure involves acquiring the structure or application of an energy storage device, identifying a first prediction model from among a plurality of first prediction models for predicting conditions other than degradation in the energy storage device that corresponds to the type of structure or application acquired, and predicting the lifespan of the energy storage device based on the predicted state value of the energy storage device predicted by the identified first prediction model and the predicted degradation value of the energy storage device predicted by a second prediction model for predicting the degradation of the energy storage device.

[0052] Typically, the lifespan of an energy storage device is predicted based on indicators related to the device's condition. Degradation of the energy storage device is a crucial indicator in lifespan prediction. For example, an energy storage device can be considered to have reached the end of its lifespan when its degradation falls below a certain level. Lifespan predictions that consider only a single indicator may not be sufficiently accurate, and it is believed that the accuracy of lifespan predictions can be further improved by combining multiple indicators. Energy storage devices are used in a variety of applications and can have a variety of configurations depending on the application. The conditions that should be considered in lifespan prediction differ depending on the configuration and application of the energy storage device. There are many values ​​that represent the condition of an energy storage device, and there are also many predictive models that predict these condition values. Appropriately selecting multiple predictive models to combine depending on the energy storage device is not easy and requires expertise.

[0053] According to the prediction method described in (18) above, the prediction accuracy can be improved by comprehensively predicting the lifespan of the energy storage device based on multiple indicators, including the predicted state value and the predicted degradation value of the energy storage device. Since the first prediction model for predicting the predicted state value, which is used together with the predicted degradation value in lifespan prediction, can be automatically identified, a first prediction model suitable for lifespan prediction can be used without operator dependence, and lifespan prediction can be performed appropriately.

[0054] (19) In the prediction method of (18) above, the first prediction model includes a model for predicting the thickness of the energy storage device and a model for predicting the reaction force of the energy storage device, and depending on the type of structure of the energy storage device, the model for predicting the thickness or the model for predicting the reaction force may be selected.

[0055] According to the configuration described in (19) above, it is possible to decide whether to select a model that predicts thickness or a model that predicts reaction force, taking into account the structure of the energy storage device. Generally, energy storage elements deform and increase in thickness in response to degradation with use. Also, the reaction force of energy storage elements increases in response to degradation with use. Depending on the structure of the energy storage device, the importance of thickness and reaction force as life prediction indicators differs. By selectively using either a model that predicts thickness or a model that predicts reaction force, depending on the structure of the energy storage device, it becomes possible to predict the lifespan by incorporating more appropriate indicators.

[0056] (20) In the prediction method described in (19) above, if an unconstrained type is obtained as the structure of the energy storage device, a model for predicting the thickness may be selected, and if a constrained type is obtained as the structure of the energy storage device, a model for predicting the reaction force may be selected.

[0057] According to the configuration described in (20) above, the thickness, which greatly affects the lifespan of an unconstrained energy storage device, and the reaction force, which greatly affects the lifespan of a constrained energy storage device, can be reflected in the lifespan prediction, so that the lifespan can be predicted with high accuracy according to the structure of the energy storage device.

[0058] (21) In any one of the prediction methods described in (18) to (20) above, the first prediction model includes a model for predicting the charge and discharge performance of different types of energy storage devices, and depending on the type of application of the energy storage device, one of the models for predicting charge and discharge performance may be selected from a plurality of models for predicting charge and discharge performance.

[0059] According to the configuration described in (21) above, it is possible to decide which model to select to predict charge and discharge performance, taking into account the application of the energy storage device. Depending on the application of the energy storage device, it is expected that the required prediction content will differ, such as what scenarios should be assumed when predicting charge and discharge performance, or what kind of predicted values ​​should be output as charge and discharge performance. By selectively using a model that predicts specific charge and discharge performance according to the application of the energy storage device, it becomes possible to predict the lifespan by incorporating more appropriate indicators.

[0060] (22) In any one of the prediction methods described in (18) to (21) above, the predicted state value of the energy storage device may include the thickness or reaction force of the energy storage device and the charge / discharge performance.

[0061] According to the configuration described in (22) above, the lifespan of the energy storage device can be accurately predicted by comprehensively evaluating the degradation of the energy storage device, as well as the thickness or reaction force of the energy storage device and its charge / discharge performance.

[0062] (23) Any one of the prediction methods described in (18) to (22) above may predict the lifespan of the energy storage device such that the predicted state value of the energy storage device satisfies predetermined requirements and the predicted degradation value of the energy storage device satisfies predetermined conditions.

[0063] According to the configuration described in (23) above, by finding a range that satisfies the predetermined conditions set for each predicted value, the lifetime can be efficiently and accurately derived through an integrated evaluation of all predictive indicators.

[0064] (24) Any one of the prediction methods described in (18) to (23) above may obtain a predicted temperature value of the energy storage device and predict the lifespan of the energy storage device such that the obtained predicted temperature value of the energy storage device, the predicted state value of the energy storage device, and the predicted degradation value of the energy storage device each satisfy predetermined conditions.

[0065] According to the configuration described in (24) above, temperature, which greatly affects the lifespan of the energy storage device, can be reflected in the lifespan prediction, thus enabling more accurate prediction of the lifespan.

[0066] (25) In any one of the prediction methods described in (18) to (24) above, a plurality of the first prediction models may be stored in advance, and a first prediction model may be selected from among the plurality of stored first prediction models, and the predicted state value of the energy storage device may be predicted using the selected first prediction model.

[0067] According to the configuration described in (25) above, by pre-storing multiple first prediction models corresponding to various energy storage device configurations and applications, it becomes possible to perform a variety of life predictions to meet various prediction requirements.

[0068] (26) A prediction device according to one aspect of the present disclosure includes a processing unit that acquires the structure or application of an energy storage device, identifies a first prediction model corresponding to the acquired type of structure or application from among a plurality of first prediction models for predicting states other than degradation in the energy storage device, and performs a process to predict the lifespan of the energy storage device based on the predicted state value of the energy storage device predicted by the identified first prediction model and the predicted degradation value of the energy storage device predicted by a second prediction model for predicting the degradation of the energy storage device.

[0069] (27) A computer program according to one aspect of the present disclosure acquires the structure or application of an energy storage device, identifies a first prediction model from among a plurality of first prediction models for predicting conditions other than degradation in the energy storage device that corresponds to the acquired type of structure or application, and causes the computer to perform a process to predict the lifespan of the energy storage device based on the predicted state value of the energy storage device predicted by the identified first prediction model and the predicted degradation value of the energy storage device predicted by a second prediction model for predicting the degradation of the energy storage device.

[0070] This disclosure will be described in detail with reference to drawings illustrating embodiments thereof.

[0071] (First Embodiment) Figure 1 is a diagram showing an overview of the prediction system 100 of this embodiment. The prediction system 100 comprises an energy storage device 1 and a prediction device 2. The energy storage device 1 and the prediction device 2 are communicated together via a network N1. A user terminal 3 is also connected to the network N1. The number of energy storage devices 1 is not limited.

[0072] The energy storage device 1 includes a rechargeable energy storage element 10, such as a secondary battery like a lead-acid battery or a lithium-ion battery. Alternatively, the energy storage element 10 may be a battery cell made of an all-solid-state battery, lead-acid battery, redox flow battery, zinc-air battery, lithium-sulfur battery, sodium-sulfur battery, silver-zinc oxide battery, or nickel-metal hydride battery, or it may be a capacitor. The energy storage device 1 may also be, for example, a module in which multiple cells are connected in series, a bank in which multiple modules are connected in series, a domain in which multiple banks are connected in parallel, or an energy storage unit including multiple domains.

[0073] The energy storage device 1 can be used, for example, in thermal power generation systems, mega solar power generation systems, wind power generation systems, uninterruptible power supplies (UPS), and stabilized power supply systems for railways. The energy storage device 1 is not limited to industrial use; it may also be for household use.

[0074] The energy storage device 1 includes measuring devices, a management device, and a communication device (not shown). The communication device may be integrated with the management device. The measuring devices repeatedly measure the current, voltage, and temperature of the energy storage element 10 at appropriate intervals. The management device acquires the measured values ​​such as current, voltage, and temperature in a time series via the communication device and manages the energy storage device 1 based on the acquired measured values ​​and values ​​calculated from those measured values ​​(e.g., SOC (State of Charge)). The management device can provide operational data, including the current, voltage, temperature, and SOC of the energy storage element 10, to the prediction device 2 via the communication device.

[0075] The prediction device 2 is an information processing device capable of various information processing and information transmission / reception related to the prediction of the state of the energy storage element 10. The prediction device 2 is, for example, a server computer, a personal computer, a quantum computer, etc. The prediction device 2 can transmit and receive information between the energy storage device 1 and the user terminal 3 via the network N1.

[0076] Network N1 is a wired or wireless network, including, for example, the Internet, a carrier network that implements wireless communication according to a predetermined mobile communication standard, or a general optical fiber line. Network N1 may also include a local network for the manufacturer or maintenance provider of the energy storage device 1.

[0077] In this embodiment, the prediction device 2 is assumed to be installed at a location separate from the energy storage device 1. Alternatively, the prediction device 2 may be installed within the facility of one of the energy storage devices 1.

[0078] User terminal 3 is an information processing terminal device used by users such as system administrators and customers, and can be a personal computer, smartphone, or tablet. User terminal 3 can receive prediction results from prediction device 2 and present them to the user. User terminal 3 may be installed within the facility of energy storage device 1.

[0079] Figure 2 is a block diagram showing an example configuration of the prediction device 2. The prediction device 2 comprises a processing unit 21, a storage unit 22, a communication unit 23, a display unit 24, and an operation unit 25.

[0080] The processing unit 21 comprises one or more processors such as a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit). The processing unit 21 includes memory, which is a temporary storage medium such as SRAM (Static Random Access Memory) or DRAM (Dynamic Random Access Memory). The processing unit 21 may also include functions such as a timer for measuring the elapsed time from the time a measurement start instruction is given until a measurement end instruction is given, a counter for counting numbers, and a clock for outputting date and time information. The processing unit 21 may be implemented in software, or part or all of it may be implemented in hardware such as an ASIC (Application Specific Integrated Circuit) or FPGA (Field Programmable Gate Array).

[0081] The storage unit 22 includes, for example, a non-volatile storage device such as a hard disk or flash memory. The storage unit 22 is separate from the prediction device 2 and may be one or more externally connected external storage devices. The storage unit 22 stores various computer programs and data referenced by the processing unit 21. In this embodiment, the storage unit 22 stores a program 221 for causing the computer to execute processing related to predicting the state of the energy storage element 10, a prediction model 222, a correction model 223, and a measurement DB (Data Base) 224. The prediction model 222 is a model for predicting the state of the energy storage element 10. The correction model 223 is a model for correcting the predicted values ​​of the prediction model 222.

[0082] The measurement DB 224 is a database that stores operational data for each energy storage element 10. The measurement DB 224 stores, for example, an energy storage element ID for identifying the energy storage element 10, the date and time of measurement of the operational data, and the operational data itself, in association with each other. The operational data includes, for example, the current, voltage, temperature, and SOC of the energy storage element 10. The prediction device 2 stores operational data in the measurement DB 224 each time it acquires it from each energy storage device 1. The measurement DB 224 may store data actually obtained not only during the operation of the energy storage element 10, but also during trial runs before operation of the energy storage element 10 or in the final design stage. The measurement DB 224 may also store data related to energy storage elements 10 other than those actually in operation, for example, data related to test energy storage elements 10 used in the construction of the prediction model 222.

[0083] A computer program (program product) including program 221 may be provided on a non-temporary recording medium 2A on which the computer program is recorded in a readable format. The recording medium 2A is a portable memory such as a CD-ROM, USB memory, or SD (Secure Digital) card. The processing unit 21 reads the desired computer program from the recording medium 2A using a reading device (not shown) and stores the read computer program in the storage unit 22. Alternatively, the computer program may be provided by communication. Program 221 may be a single computer program or may consist of multiple computer programs. Program 221 may also be executed on a single computer or executed collaboratively by multiple computers.

[0084] The communication unit 23 includes a communication device that enables communication via the network N1. The processing unit 21 sends and receives data between the energy storage device 1 and the user terminal 3 via the communication unit 23.

[0085] The display unit 24 includes a display device such as a liquid crystal display or an organic electroluminescent (OLED) display. The display unit 24 displays various information according to instructions from the processing unit 21.

[0086] The operation unit 25 is an interface that receives user input. The operation unit 25 includes, for example, a keyboard, mouse, touch panel device with a built-in display, speaker, and microphone. The operation unit 25 receives user input and sends control signals to the processing unit 21 according to the content of the operation.

[0087] The prediction device 2 may be configured to receive operations via an externally connected computer and output information to be notified to the external computer. In this case, the prediction device 2 does not need to have a display unit 24 and an operation unit 25.

[0088] Figure 3 illustrates the method used by the prediction device 2 to predict the state of the energy storage element 10. The prediction device 2 uses a prediction model 222 and a correction model 223 stored in the memory unit 22 to predict the state of the energy storage element 10.

[0089] The prediction model 222 is a model that predicts the state of the energy storage element 10. The prediction model 222 is a model that mathematically describes the state of the energy storage element using calculation formulas such as algebraic equations and differential equations, and parameters. Examples of the state of the energy storage element 10 predicted by the prediction model 222 include the degradation state of the energy storage element 10 (e.g., energy storage capacity, state of health (SOH), internal resistance, charge / discharge characteristics, or the amount of decrease thereof), charge / discharge capacity (SFO), thickness, reaction force, etc. In this embodiment, the energy storage capacity is predicted using the prediction model 222.

[0090] As an example, the prediction model 222 takes usage history data representing the usage history of the energy storage element 10 as input and outputs a predicted value for the degradation state of the energy storage element 10. By providing the prediction model 222 with usage history data over the prediction period, it is possible to predict the trend of the predicted value for the degradation state over the prediction period. Figure 3 shows the trend of the degradation state output by the prediction model 222 in graph format. The horizontal axis of the graph represents the number of years of operation, and the vertical axis represents the degradation state (for example, full charge capacity).

[0091] Usage history data represents the changes in the usage status of the energy storage element during the prediction period (for example, from time t1 to time tn). Usage history data includes, for example, temperature change data showing the change in the temperature T of the energy storage element, and SOC change data showing the change in SOC. Usage history data may also include data showing changes in the power of the energy storage element, elapsed time since the start of use, and the number of cycles since the start of use. Usage history data can be derived based on measured values ​​such as current, voltage, and temperature of the energy storage element stored in the measurement DB224. The prediction model may also be input with assumed usage history data corresponding to assumed measured values ​​for the prediction period predicted from actual operational data.

[0092] The prediction model 222 calculates the degradation value Qdeg of the energy storage element 10 after a predetermined period of time according to the following formula and predicts the energy storage capacity: Qdeg = Qcnd + Qcur

[0093] Here, Qcnd is the non-energized degradation value, and Qcur is the energized degradation value. The non-energized degradation value Qcnd can be calculated, for example, by Qcnd = K1 × √(t). The energized degradation value Qcur can be calculated, for example, by Qcur = K2 × √(t). K1 and K2 are degradation coefficients, respectively. The degradation coefficient is a parameter that represents degradation and is a function of SOC and temperature T. If the storage capacity at time t1 is known, the storage capacity at time tn can be calculated based on the degradation value Qdeg. If the storage capacity at time t1 is Qt1 and the storage capacity at time tn is Qtn, then the predicted value of the storage capacity can be obtained by Qtn = Qt1 - Qdeg.

[0094] The prediction model 222 can be constructed, for example, by collecting test data from a durability test using a large number of test cells over a predetermined test period, and determining the exponents and parameters of the roots. The structure of the prediction model 222 is not limited to the example described above, and a mathematical model including appropriate formulas and parameters can be adopted.

[0095] The correction model 223 is a model that corrects the error between the predicted value of the energy storage capacity predicted by the prediction model 222 and the measured value of the energy storage capacity. The correction model 223 is a learning model generated by machine learning. The correction model 223 is intended to be used as a program module that constitutes part of artificial intelligence software. The correction model 223 is a model that takes usage history data representing the usage history of the energy storage element 10 as input and outputs a correction value corresponding to said usage history data. The usage history data includes, for example, data showing the changes in the temperature T, SOC, power, elapsed time since the start of use, and the number of cycles since the start of use of the energy storage element 10 during the prediction period related to the predicted value to be corrected. The content of the usage history data input to the correction model 223 may be the same as or different from the usage history data input to the prediction model 222.

[0096] The correction value output from the correction model 223 is a value used to correct the error between the predicted value from the prediction model 222 and the measured value. The "measured value" is the actual value measured when the energy storage system is actually used. When the prediction model predicts the energy storage capacity at a future point in time, the error between the predicted value and the measured value refers to the error between the predicted value and the assumed measured value at that future point in time.

[0097] The configuration of the correction model 223 is not particularly limited and can be constructed using any learning algorithm such as supervised learning or unsupervised learning. The correction model 223 may be, for example, a neural network, RNN (Recurrent Neural Network), CNN (Convolutional Neural Network), Transformer, support vector machine, logistic regression, decision tree, XGBooster (eXtreme Gradient Boosting), etc.

[0098] The correction model 223 can be generated, for example, by preparing a large amount of training data in which data indicating known correction values ​​at a specific time are labeled to the usage history data of the energy storage element 10 at that specific time, and then training an untrained correction model using this training data. The correct correction value is the comparison between the predicted value of the energy storage capacity by the prediction model 222 at the specific time (i.e., the same time point) and the actual measured value. The correction value is, for example, the difference between the predicted value and the measured value. Alternatively, the correction value may be the ratio or percentage between the predicted value and the measured value.

[0099] Training data can be generated by collecting the predicted values ​​of a prediction model 222 for a given energy storage element over a specified period, and the measured values ​​obtained when charging and discharging are actually performed during that period. The energy storage element used to generate the training data may be, for example, a simulated cell that mimics an energy storage element in a system, a test cell, or a market product cell. The training data may include correction values ​​based on the measured values ​​during operation of the energy storage element 10 being predicted and the predicted values ​​of the prediction model 222. The training data may be weighted according to the content of the usage history data. For example, a larger weight may be assigned to each training data point if the similarity between the usage history data and the usage history data of the energy storage element 10 corresponding to the corrected predicted value is high.

[0100] The prediction device 2 uses the obtained training data to train the correction model 223. The prediction device 2 inputs the usage history data included in the training data into the correction model 223 and obtains the correction value output from the correction model 223 after intermediate calculation processing. The prediction device 2 compares the correction value output from the correction model 223 with the correction value included in the training data and optimizes the parameters (e.g., weights) of the correction model 223 so that the correction value output from the correction model 223 approaches the correct value. The method for optimizing the parameters is not particularly limited, but for example, backpropagation or gradient descent may be used.

[0101] The correction model 223 is not limited to one generated by the prediction device 2. The prediction device 2 may acquire a trained correction model 223 generated on an external server and store it in the storage unit 22. The correction model 223 may be generated or pre-trained on an external server and then trained or fine-tuned in the prediction device 2.

[0102] The correction model 223 is not limited to machine learning models. The correction model 223 may, for example, calculate a correction value according to the usage history data using a rule-based method.

[0103] Multiple types of prediction model 222 and correction model 223 may be provided, depending on the configuration, application, and calculation conditions of the energy storage element 10.

[0104] The prediction device 2 inputs the usage history data of the energy storage element 10 to be predicted into the prediction model 222 and obtains a predicted value of the energy storage capacity output from the prediction model 222. The prediction device 2 also inputs the usage history data of the energy storage element 10 into the correction model 223 and obtains a correction value output from the correction model 223. The prediction device 2 corrects the predicted value of the energy storage capacity using the obtained correction value. The prediction device 2 corrects the predicted value, for example, by adding or subtracting the correction value to the predicted value of the energy storage capacity. As shown in Figure 3, by applying the correction to the predicted value of the energy storage capacity, which is represented by the dashed line, the corrected predicted value of the energy storage capacity, which is represented by the solid line, is obtained. Based on the correction value corresponding to each point in time during the prediction period, the energy storage capacity at each point in time is corrected accordingly. Through the above process, the predicted value of the energy storage capacity, which includes errors, can be appropriately corrected.

[0105] Figure 4 is a flowchart showing an example of a processing procedure performed by the prediction device 2. The processing in each of the following flowcharts is performed by the processing unit 21 according to the program 221 stored in the memory unit 22 of the prediction device 2. The prediction device 2 repeatedly performs the following processing, for example, at regular intervals or when it receives a request for life prediction from the user terminal 3.

[0106] The processing unit 21 of the prediction device 2 acquires usage history data of the energy storage element 10 to be predicted based on the operation data of the energy storage element 10 to be predicted stored in the measurement DB 224 (step S11). The processing unit 21 uses the prediction model 222 to acquire a predicted value of the energy storage capacity according to the derived usage history data (step S12). The processing unit 21 uses the correction model 223 to estimate a correction value according to the usage history data (step S13). If multiple types of prediction models 222 and correction models 223 are stored, the processing unit 21 may select the prediction model 222 and correction model 223 to be used in steps S12 and S13 according to the configuration, application and calculation conditions of the energy storage element 10 to be predicted.

[0107] The processing unit 21 corrects the predicted value of the energy storage capacity by the prediction model 222 using the estimated correction value (step S14). The processing unit 21 outputs the prediction result representing the corrected predicted value of the energy storage capacity to the user terminal 3 (step S15), and the series of processes ends.

[0108] The main entity performing each process in the flowchart above is not limited to the prediction device 2; some or all of the above processes may be executed by, for example, the user terminal 3, a simulator, or another computer. The prediction device 2 may cooperate with these other devices to execute a series of processes. The prediction model 222 may be stored in another device, and the prediction device 2 may obtain the predicted value of the energy storage capacity obtained from the simulation by the other device via communication. The output destination of the prediction results is not limited to the user terminal 3; for example, it may be the display unit 24.

[0109] (Second Embodiment) In the second embodiment, a configuration is described in which the prediction model 222 is updated in addition to the correction by the correction model 223. The following mainly describes the differences from the first embodiment, and components common to the first embodiment are denoted by the same reference numerals and their detailed descriptions are omitted.

[0110] Depending on the magnitude of the difference between the energy storage element 10 to be predicted and the energy storage element 10 used for testing, and the quality of the data used to construct the correction model 223, it may not be possible to obtain a prediction with sufficient accuracy by simply adding the correction value from the correction model 223. Furthermore, it is expected that the accuracy of the correction model 223 itself will decrease as the energy storage element 10 deteriorates or changes in its usage conditions occur. In this embodiment, further improvements in prediction accuracy are made by updating the correction model 223 and the prediction model 222.

[0111] Figures 5 and 6 are flowcharts showing an example of the processing procedure performed by the prediction device 2 of the second embodiment. The prediction device 2 performs the following processing at appropriate intervals, in parallel with the processing shown in the flowchart in Figure 4, for example.

[0112] The processing unit 21 of the prediction device 2 determines whether or not the correction model 223 needs to be updated (step S21). The processing unit 21 determines that the correction model 223 needs to be updated if, for example, a pre-set update event is detected, and determines that the correction model 223 does not need to be updated if no update event is detected. Update events include, for example, when the difference between the predicted value of the energy storage capacity of the energy storage element 10 by the prediction model 222 and the measured value of the energy storage capacity based on operational data is greater than or equal to a pre-set predetermined value, or when it is the update timing. Alternatively, the processing unit 21 may proceed to step S22 without performing the determination in step S21.

[0113] If it is determined that updating the correction model 223 is unnecessary (S21: YES), the processing unit 21 terminates the process.

[0114] If it is determined that the correction model 223 needs to be updated (S21: NO), the processing unit 21 updates the correction model 223 (step S22). The processing unit 21 updates the correction model 223 by, for example, retraining the correction model 223 using the learning method described above, with new training data that includes a different dataset from the previously learned training data, either in place of or in addition to the previously learned training data. The new training data can be generated, for example, based on test data with a different test period than the test data included in the previously learned training data.

[0115] The processing unit 21 performs the same processing as in steps S12 to S14 of the first embodiment to obtain a first predicted value of the energy storage capacity using the prediction model 222 (step S23), estimate a first correction value for the first predicted value using the updated correction model 223 (step S24), and correct the first predicted value using the first correction value (step S25).

[0116] The processing unit 21 calculates a first error between the corrected first predicted value of the energy storage element 10 and the measured value of the energy storage capacity (step S26). The first error is calculated, for example, as the average value of the difference between the corrected first predicted value and the measured value over the most recent predetermined period, the difference in energy storage capacity at the point when the expected number of years (i.e., product life) is reached, which is determined from the trend of the corrected first predicted value and the trend of the measured value, and the difference in the period until the above expected number of years is reached. The measured value of the energy storage capacity is obtained, for example, by collecting operational data of the energy storage element 10 over a specific period after prediction by the prediction model 222.

[0117] The processing unit 21 determines whether the first error between the calculated corrected first predicted value and the measured value is within a preset first tolerance range (step S27). The first tolerance range is defined by an upper limit, a lower limit, or a combination thereof. The second and third tolerance ranges, described later, are defined similarly. The first to third tolerance ranges may be the same range or may be different ranges. If the processing unit 21 determines that the first error between the corrected first predicted value and the measured value is within the first tolerance range (S27: YES), the processing unit 21 terminates the process.

[0118] If the first error between the corrected first predicted value and the measured value is determined to be outside the first acceptable range (S27: NO), the processing unit 21 updates the parameters in the prediction model 222 (step S28). The processing unit 21 updates the parameters in the prediction model 222 by, for example, readjusting the parameters using new test data relating to different test periods and test conditions than the previous test data, either in place of or in addition to the test data used to update the prediction model 222 in the previous update.

[0119] The processing unit 21 obtains a second predicted value of the energy storage capacity predicted by the prediction model 222 after the parameters have been updated (step S29). The processing unit 21 calculates a second error between the obtained second predicted value and the measured value of the energy storage capacity (step S30). The calculation of the second error may be the same as that of the first error.

[0120] The processing unit 21 determines whether the second error between the second predicted value and the measured value is within a preset second tolerance range (step S31). If it determines that the second error between the second predicted value and the measured value is within the second tolerance range (S31: YES), the processing unit 21 terminates the process.

[0121] If the second error between the second predicted value and the measured value is determined to be outside the second acceptable range (S31: NO), the processing unit 21 updates the correction model 223 (step S32). The processing unit 21 updates the correction model 223 by retraining the correction model 223 using new training data that includes a new correction value which is the difference between the second predicted value and the measured value obtained by the prediction model 222 after the parameters have been updated.

[0122] The processing unit 21 estimates a second correction value for the second predicted value using the updated correction model 223 (step S33). The processing unit 21 corrects the second predicted value using the estimated second correction value (step S34). The processing unit 21 calculates a third error between the corrected second predicted value and the measured value of the energy storage capacity (step S35). The calculation of the third error is similar to that of the first error.

[0123] The processing unit 21 determines whether the third error between the corrected second predicted value and the measured value is within a preset third tolerance range (step S36). If it determines that the third error between the corrected second predicted value and the measured value is within the third tolerance range (S36: YES), the processing unit 21 terminates the process.

[0124] If the processing unit 21 determines that the third error between the corrected second predicted value and the measured value is outside the third acceptable range (S36: NO), it updates the calculation formula in the prediction model 222 (step S37). The processing unit 21 updates the calculation formula in the prediction model 222 by, for example, resetting the calculation formula based on the new test data used to update the parameters. The processing unit 21 may also update the prediction model 222 in the storage unit 22 by accepting the calculation formula that has been manually reset. The processing unit 21 then terminates the series of processes.

[0125] Subsequently, the processing unit 21 executes the flowchart in Figure 4 using the updated correction model 223 and prediction model 222 as needed to predict the energy storage capacity.

[0126] The processing unit 21 may determine an abnormal state of the energy storage element 10 based on the update status and content of the correction model 223 and the prediction model 222. For example, after step S37, the processing unit 21 continues to perform a process to determine the possibility of an abnormality.

[0127] The processing unit 21 determines whether the update frequency of the correction model 223 and the prediction model 222 conforms to a preset abnormal condition. For example, if it determines that the update frequency is too low and therefore does not conform to the abnormal condition, the processing unit 21 terminates processing without outputting abnormal information. For example, if it determines that the update frequency is too high and therefore conforms to the abnormal condition, the processing unit 21 outputs abnormal information indicating that the energy storage element 10 conforms to the abnormal condition to a predetermined output destination. The output destination for the abnormal information may include the display unit 24, the user terminal 3, etc. If an abnormality notification is received at the output destination, more detailed abnormality detection processing may be performed. The abnormality information may include an instruction for automatic execution of the abnormality detection processing.

[0128] The determination of whether or not an abnormal condition is met is not limited to the high update frequency described above, but may also be based on the direction of correction of the predicted value by the updated prediction model 222, the direction of correction of the predicted value based on the correction value by the updated correction model 223, etc. For example, if the direction of correction of the predicted value when using the updated model is in the direction of increasing deterioration, it can be determined that an abnormal condition is met. If the amount of correction of the predicted value or the correction value in the direction of increasing deterioration is greater than or equal to a predetermined value, it may be determined that an abnormal condition is met. If the update frequency is greater than or equal to a predetermined number of times or less than or equal to a predetermined interval, and the direction of correction of the predicted value is in the direction of increasing deterioration, it may be determined that an abnormal condition is met.

[0129] (Third Embodiment) Figure 7 shows an overview of the energy storage information processing system 200 according to the third embodiment. The energy storage information processing system 200 comprises an information processing device 4 and operator devices 5a, 5b, and 5c. The information processing device 4 and each of the operator devices 5a to 5c are connected to each other via a network N2 such as the Internet.

[0130] The information processing device 4 is a device capable of various information processing and information transmission and reception, such as a server computer, personal computer, or quantum computer. The information processing device 4 is used by business operator X, which is engaged in the business of designing, introducing, operating, and maintaining energy storage systems 6x. The information processing device 4 has an operation DB (Data Base) 41 for storing operational data of the energy storage system 6x, a prediction model 42, and a correction model 43. The number of energy storage systems 6x owned by business operator X may be two or more.

[0131] The operator's equipment 5a is a device capable of various information processing and information transmission / reception, such as a server computer, personal computer, or quantum computer. Operator's equipment 5a is used by operator A, which is engaged in the business of designing, introducing, operating, and maintaining the energy storage system 6a. Operator's equipment 5a has an operation DB 51a for storing operation data of the energy storage system 6a, and a prediction model 52a. Similarly, operator's equipment 5b is used by operator B of the energy storage system 6b, and has an operation DB 51b for storing operation data of the energy storage system 6b, and a prediction model 52b. Operator's equipment 5c is used by operator C of the energy storage system 6c, and has an operation DB 51c for storing operation data of the energy storage system 6c, and a prediction model 52c. The number of energy storage systems 6a to 6c owned by each operator A to C may be two or more.

[0132] In the following explanation, if it is not necessary to distinguish between energy storage systems 6a to 6c and 6x, they will simply be referred to as energy storage system 6. If it is not necessary to distinguish between operator equipment 5a to 5c, they will simply be referred to as operator equipment 5. If it is not necessary to distinguish between operation databases 51a to 51c, they will simply be referred to as operation database 51. If it is not necessary to distinguish between prediction models 52a to 52c, they will simply be referred to as prediction model 52.

[0133] Energy storage systems 6a-6c and 6x are used, for example, in thermal power generation systems, mega solar power generation systems, wind power generation systems and battery storage equipment such as ESS (Energy Storage System), V2X (Vehicle to Load, Home, Grid, etc.), including power conditioners with battery storage equipment and power conditioners for electric vehicles, uninterruptible power supplies (UPS), and stabilized power supply systems for railways. Energy storage systems 6a-6c and 6x are not limited to industrial use, but may also be for household use.

[0134] The energy storage system 6 comprises multiple rechargeable energy storage elements, such as lead-acid batteries or lithium-ion batteries, which are secondary batteries. Alternatively, the energy storage elements may be battery cells such as all-solid-state batteries, lead-acid batteries, redox flow batteries, zinc-air batteries, lithium-sulfur batteries, sodium-sulfur batteries, silver-zinc oxide batteries, and nickel-metal hydride batteries, or they may be capacitors. The energy storage system 6 may also be, for example, a module in which multiple cells are connected in series, a bank in which multiple modules are connected in series, a domain in which multiple banks are connected in parallel, or an energy storage unit containing multiple domains.

[0135] The energy storage system 6 includes, for example, measuring devices, management devices, and communication devices (not shown). The communication device may be integrated with the management device. The communication device is connected to any of the corresponding information processing devices 4 and operator devices 5a to 5c via a wired or wireless network (not shown). The communication device may be connected to the information processing device 4 or operator devices 5a to 5c via network N2.

[0136] The management device acquires measured values ​​such as current, voltage, and temperature of the energy storage elements measured by the measuring device, and manages the energy storage system 6 based on the acquired measured values ​​and values ​​calculated from those measured values ​​(for example, SOC (State of Charge)). Each measured value is measured repeatedly at appropriate intervals. The measured values ​​such as current, voltage, temperature, and SOC of the energy storage elements are provided from the management device to the information processing device 4 or the operator devices 5a to 5c, and are stored as operation data in the operation databases 41 and 51. The operation data may include not only data obtained during the operation of the energy storage system 6, but also data actually obtained during trial runs before operation of the energy storage system 6 and during the final design stage.

[0137] Prediction models 42, 52a to 52c are models that predict the state of an energy storage element. Prediction models 42, 52a to 52c are models that mathematically describe the state of an energy storage element using calculation formulas such as algebraic equations and differential equations, and parameters. Examples of the state of an energy storage element predicted by prediction models 42, 52a to 52c include the degradation state of the energy storage element (e.g., energy storage capacity, state of health (SOH), internal resistance, charge / discharge characteristics, or the amount of decrease thereof), charge / discharge capacity (SFO), thickness, reaction force, etc. In this embodiment, prediction models 42, 52a to 52c are assumed to be models that predict the degradation state. Prediction models 42, 52a to 52c may employ appropriate prediction models constructed by known prediction methods. Prediction models 42, 52a to 52c may be different from each other, or they may include the same model.

[0138] Correction model 43 is a model that corrects the error between the predicted value of the degradation state predicted by prediction models 42, 52a to 52c and the measured value of the energy storage capacity. "Measured value" refers to the actual value measured when the energy storage element is actually used. When predicting the degradation state at a future point in time using prediction models 42 and 52, the error between the predicted value and the measured value refers to the error between the predicted value and the assumed measured value at a future point in time.

[0139] The information processing device 4 uses the correction model 43 to implement a prediction value correction service that provides correction information to businesses A to C, other than its own business X, regarding the correction of prediction values ​​of the deterioration state based on prediction models 52a to 52c held by businesses A to C. Business X corresponds to the first business, and businesses A to C correspond to the second business. The correction information may be a correction value for the prediction value, or it may be a prediction value corrected based on the correction value (i.e., a corrected prediction value).

[0140] Figure 8 is a block diagram showing an example configuration of the information processing device 4. The information processing device 4 comprises a processing unit 401, a storage unit 402, a communication unit 403, a display unit 404, and an operation unit 405.

[0141] The processing unit 401 comprises one or more processors such as a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit). The processing unit 401 includes memory, which is a temporary storage medium such as SRAM (Static Random Access Memory) or DRAM (Dynamic Random Access Memory). The processing unit 401 may also include functions such as a timer for measuring the elapsed time from the time a measurement start instruction is given to the time a measurement end instruction is given, a counter for counting numbers, and a clock for outputting date and time information. The processing unit 401 may be implemented in software, or part or all of it may be implemented in hardware such as an ASIC (Application Specific Integrated Circuit) or FPGA (Field Programmable Gate Array).

[0142] The storage unit 402 includes, for example, a non-volatile storage device such as a hard disk or flash memory. The storage unit 402 is separate from the information processing device 4 and may be one or more externally connected external storage devices. The storage unit 402 stores various computer programs and data referenced by the processing unit 401. In this embodiment, in addition to the operation DB 41, prediction model 42, and correction model 43 described above, the storage unit 402 stores a program 44 that causes the computer to execute processing related to correcting the predicted value of the degradation state. The storage unit 402 may store multiple correction models 43.

[0143] The operation database 41 stores, for example, for each of the multiple energy storage elements included in the energy storage system 6x, the energy storage element ID for identifying the energy storage element, the date and time of measurement of the operation data, and the operation data itself, in a time-series manner, with associated information. The operation data is information that represents the operating status of the energy storage element and includes, for example, the current, voltage, temperature, and SOC of the energy storage element. The information processing device 4 stores the operation data in the operation database 41 each time it acquires operation data from the energy storage system 6x.

[0144] A computer program (program product) including program 44 may be provided on a non-temporary recording medium 4A on which the computer program is recorded in a readable format. The recording medium 4A is a portable memory such as a CD-ROM, USB memory, or SD (Secure Digital) card. The processing unit 401 reads the desired computer program from the recording medium 4A using a reading device (not shown) and stores the read computer program in the storage unit 402. Alternatively, the computer program may be provided by communication. Program 44 may be a single computer program or may consist of multiple computer programs. Program 44 may also be executed on a single computer or may be executed collaboratively by multiple computers.

[0145] The communication unit 403 is equipped with a communication device that enables communication via the network N2. The processing unit 401 transmits and receives data with each operator's equipment 5a to 5c via the communication unit 403.

[0146] The display unit 404 includes a display device such as a liquid crystal display or an organic EL (Electro-Luminescence) display. The display unit 404 displays various information according to instructions from the processing unit 401.

[0147] The operation unit 405 is an interface that receives user input. The operation unit 405 includes, for example, a keyboard, mouse, touch panel device with a built-in display, speaker, and microphone. The operation unit 405 receives user input and sends control signals to the processing unit 401 according to the content of the operation.

[0148] The information processing device 4 may be configured to receive operations via an externally connected computer and output information to be notified to the external computer. In this case, the information processing device 4 does not need to include a display unit 404 and an operation unit 405.

[0149] Figure 9 illustrates the method for correcting predicted values ​​of the deterioration state using correction model 43. After explaining the outlines of prediction models 42, 52a to 52c and correction model 43 using Figure 9, the method for correcting predicted values ​​using each model will be explained. In the following, the method for correcting predicted values ​​will be explained using the case of correcting predicted values ​​using prediction model 42 as an example.

[0150] As an example, the prediction model 42 takes operational data of the energy storage element as input and outputs a predicted value for the degradation state of the energy storage element. By providing the prediction model 42 with time-series operational data over the prediction period (for example, from time t1 to time tn), it is possible to predict the trend of the predicted value for the degradation state over the prediction period. Figure 9 shows the trend of the degradation state output by the prediction model 42 in graph format. The horizontal axis of the graph represents the number of years of operation, and the vertical axis represents the energy storage capacity, which is an example of a degradation state.

[0151] The operational data input to the prediction model 42 may include, for example, SOC trend data showing the trend of SOC calculated based on the current and voltage of the energy storage element during the prediction period, and temperature trend data showing the trend of the temperature of the energy storage element. The operational data input to the prediction model 42 may also include power trend data showing the trend of the power of the energy storage element, elapsed time trend data showing the trend of the elapsed time since the start of use, cycle count trend data showing the trend of the number of cycles since the start of use, etc. The operational data input to the prediction model 42 may also be assumed operational data corresponding to assumed measured values ​​predicted from actual operational data.

[0152] The prediction model 42 calculates the degradation value QCeg of the energy storage element after a predetermined period of time according to the following formula and predicts the degradation state: QCeg = QBnC + QBu

[0153] Here, QBnC is the degradation value when not energized, and QBur is the degradation value when energized. The degradation value when not energized, QBnC, can be calculated, for example, by QBnC = K1 × √(t). The degradation value when energized, QBur, can be calculated, for example, by QBur = K2 × √(t). K1 and K2 are degradation coefficients, respectively. The degradation coefficient is a parameter that represents degradation and is a function of SOC and temperature T. If the degradation state at time t1 is known, the degradation state at time tn can be determined based on the degradation value QCeg. If the degradation state at time t1 is Qt1 and the degradation state at time tn is Qtn, then the degradation state can be estimated by Qtn = Qt1 - QCeg.

[0154] The prediction model 42 can be constructed, for example, by collecting test data from a durability test using a large number of test cells over a predetermined test period, and determining the exponents and parameters of the roots. The structure of the prediction model 42 is not limited to the example described above, and may be a mathematical model including appropriate formulas and parameters.

[0155] Similarly, prediction models 52a to 52c are configured to output predicted values ​​for the deterioration state according to operational data. The configurations of prediction models 52a to 52c are not limited to the example described above and may differ. Prediction models 42, 52a to 52c may also be learning models constructed using machine learning.

[0156] The correction model 43 is a model that takes operational data of an energy storage element as input and outputs a correction value corresponding to that operational data. The correction model 43 is a learning model generated by machine learning. The correction model 43 is intended to be used as a program module that constitutes part of artificial intelligence software.

[0157] The operational data input to the correction model 43 includes, for example, SOC trend data, temperature trend data, elapsed time trend data, and cycle number trend data of the energy storage element during the forecast period. The content of the operational data input to the correction model 43 may be partially or entirely the same as the operational data input to the prediction model 42, or it may be different.

[0158] The correction value output from the correction model 43 is a value used to correct the error between the predicted value from the prediction model 42 and the measured value. The "measured value" is the actual value measured when the energy storage system 6 is actually used. When the prediction model 42 predicts the degradation state at a future point in time, the error between the predicted value and the measured value means the error between the predicted value and the assumed measured value at a future point in time.

[0159] The configuration of the correction model 43 is not particularly limited and can be constructed using any learning algorithm such as supervised learning or unsupervised learning. The correction model 43 may be, for example, a neural network, RNN (ReBurrent NeurXl Network), BNN (Bonvolution NeurXl Network), TrXnsformer, support vector machine, logistic regression, decision tree, XGAoost (eXtreme GrXCient Aoosting), etc.

[0160] The correction model 43 can be constructed, for example, by preparing a large amount of training data in which data indicating known correction values ​​at a specific time are labeled to the operational data of the energy storage element at that specific time, and then training an untrained correction model 43 using this training data. The correct correction value is the error between the predicted value of the degradation state by the prediction model 42 at the specific time (i.e., the same time) and the actual measured value. The correction value is, for example, the difference between the predicted value and the measured value. Alternatively, the correction value may be the ratio or percentage between the predicted value and the measured value.

[0161] Training data can be generated by collecting predicted values ​​from a prediction model 42 for a given period, and measured values ​​obtained when charging and discharging are actually performed during the prediction period for a given energy storage element. Examples of energy storage elements used to generate training data include simulated cells that mimic energy storage elements in a system, test cells, and market product cells. The training data may include correction values ​​based on measured values ​​during operation of the target energy storage element and the predicted values ​​from the prediction model 42.

[0162] The training data may be weighted according to the content of the operational data. For example, the higher the similarity between the operational data in each training data set and the operational data of the energy storage element corresponding to the predicted value to be corrected, the larger the weight may be set. The similarity may be calculated using Euclidean distance or cosine similarity based on the vector of operational data. If the predicted value is in a degraded state, it is preferable to increase the weight assigned to the training data the higher the similarity between the current and voltage or power trends of the energy storage element.

[0163] The information processing device 4 uses the obtained training data to train the correction model 43. The information processing device 4 inputs the operational data included in the training data into the correction model 43 and obtains the correction value output from the correction model 43 after intermediate calculation processing. The information processing device 4 compares the correction value output from the correction model 43 with the correction value included in the training data and optimizes the parameters (e.g., weights) of the correction model 43 so that the correction value output from the correction model 43 approaches the correct value. The method for optimizing the parameters is not particularly limited, but for example, backpropagation or gradient descent may be used.

[0164] The correction model 43 is not limited to one generated by the information processing device 4. The information processing device 4 may acquire a trained correction model 43 generated on an external server and store it in the storage unit 402. The correction model 43 may be generated or pre-trained on an external server and then trained or fine-tuned on the information processing device 4.

[0165] The correction model 43 is not limited to machine learning models. The correction model 43 may, for example, calculate correction values ​​according to operational data using a rule-based method.

[0166] Multiple types of correction models 43 may be provided depending on the configuration, application, and calculation conditions of the energy storage element.

[0167] The predicted values ​​of the prediction model 42 are corrected using the correction values ​​obtained by the correction model 43 described above. For example, the predicted values ​​are corrected by adding or subtracting the correction values ​​to the predicted values ​​from the prediction model 42. As shown in Figure 9, by applying corrections to the predicted values ​​of the deterioration state, which are represented by the dashed line, the corrected predicted values ​​of the deterioration state, which are represented by the solid line, are obtained. Based on the correction values ​​corresponding to each point in time during the prediction period, the deterioration state at each point in time is corrected accordingly. Through the above process, the predicted values ​​of the deterioration state, including errors, can be appropriately corrected.

[0168] The information processing device 4 uses the correction model 43 described above to perform correction processing on the predicted values ​​using its own prediction model 42. The information processing device 4 also performs correction processing on the predicted values ​​using the prediction model 52 received from the business operator device 5. An example of correcting the predicted values ​​using the prediction model 52a obtained from the business operator device 5a will be described below.

[0169] When using the correction service, business operator A uses business operator equipment 5a to associate the operational data of the target energy storage element in the energy storage system 6a with the predicted value obtained from its own prediction model 52a, and transmits it to the information processing device 4.

[0170] The information processing device 4 may obtain the prediction model 52a from the operator's equipment 5a instead of the prediction values ​​of the prediction model 52a. If the prediction model 52a is obtained, the information processing device 4 provides the obtained prediction model 52a with the operation data of the target energy storage element and obtains the prediction values ​​output from the prediction model 52a.

[0171] The information processing device 4 inputs the acquired operational data of the target energy storage element into the correction model 43 and obtains the correction value output from the correction model 43.

[0172] The information processing device 4 corrects the predicted values ​​of the deterioration state by the prediction model 52a using the acquired correction values. The information processing device 4 transmits the prediction result, which represents the corrected predicted values ​​of the deterioration state, to the operator's device 5a. Alternatively, the prediction result may be the correction values ​​for each predicted value.

[0173] The information processing device 4 can correct the predicted values ​​from each of the prediction models 52a to 52c by performing the above-described process on each of the predicted values ​​from each of the prediction models 52a to 52c that it receives from each of the business operator devices 5a to 5c.

[0174] The information processing device 4 may reconstruct the correction model 43 in accordance with the prediction model 52 used to generate the predicted values ​​to be corrected. Reconstruction of the correction model 43 can be performed by collecting the error between the predicted values ​​from the prediction model 52 and the actual measured values, and then retraining using new training data that includes the collected error.

[0175] Prior to correcting the predicted values ​​using the prediction model 52a, for example, the information processing device 4 acquires operational data relating to one or more energy storage elements, the predicted values ​​of the prediction model 52a, and the error between the predicted values ​​and the actual measured values ​​from the operator's equipment 5a, relating them together. The information processing device 4 may acquire the time-synchronized operational data, predicted values, and error values ​​as time-series data. The error between the predicted values ​​and the measured values ​​is the same as the corrected values ​​output by the correction model 43, and may be, for example, the difference between the predicted values ​​and the measured values ​​at the same point in time, or it may be a ratio or percentage. The measured values ​​to be compared with the predicted values ​​are obtained, for example, from the results of durability tests conducted by operator A.

[0176] The calculation of the error may be performed by the information processing device 4. The information processing device 4 receives the predicted values ​​of the prediction model 52a and the measured values ​​from the operator's device 5a, instead of the predicted values ​​of the prediction model 52a and the error between the predicted values ​​and the measured values. The information processing device 4 can calculate the error by comparing the received predicted values ​​and measured values.

[0177] The information processing device 4 generates new training data by associating operational data acquired from the operator's equipment 5a with correction values ​​indicating the error between the predicted values ​​and measured values ​​from the prediction model 52a. The information processing device 4 updates the correction model 43 by retraining it using the newly generated training data in the learning method described above, either in place of or in addition to the previously learned training data. The correction model 43 is reconstructed as a result of the update of the correction model 43 for the prediction model 52a. The information processing device 4 stores the reconstructed correction model 43 for the prediction model 52a in the storage unit 402, associating it with the identification information of the operator's equipment 5a. Alternatively, the correction model 43 for the prediction model 52a may be stored as associated with the identification information of the operator A that owns the operator's equipment 5a, or the identification information of the prediction model 52a used to generate the predicted values.

[0178] Similarly, the information processing device 4 acquires operational data relating to one or more energy storage elements, predicted values ​​from the prediction model 52b, and the error between the predicted values ​​and measured values ​​of the degradation state from the operator's equipment 5b, relating them together. The information processing device 4 reconstructs the correction model 43 for the prediction model 52b by updating the correction model 43 based on the acquired operational data and the error between the predicted values ​​and measured values ​​of the prediction model 52b. The information processing device 4 acquires operational data relating to one or more energy storage elements relating to one or more energy storage elements, predicted values ​​from the prediction model 52c, and the error between the predicted values ​​and measured values ​​of the degradation state from the operator's equipment 5c, relating them together. The information processing device 4 reconstructs the correction model 43 for the prediction model 52c by updating the correction model 43 based on the acquired operational data and the error between the predicted values ​​and measured values ​​of the prediction model 52c.

[0179] By fine-tuning the correction model 43 for each of the 52 types of prediction models based on the correspondence between operational data and the actual error between predicted and measured values, the accuracy of prediction error correction can be improved for each of the prediction models 52a to 52c held by each of the operators A to C.

[0180] If correction models 43 are prepared for each of the prediction models 52a to 52c, the information processing device 4 corrects the prediction values ​​using the correction model 43 corresponding to the prediction model 52a to 52c used to generate the prediction values ​​to be corrected. For example, if the information processing device 4 obtains prediction values ​​from the operator's equipment 5a using prediction model 52a, it identifies the correction model 43 for prediction model 52a that corresponds to the identification information of the operator's equipment 5a from among the multiple correction models 43 stored in the storage unit 402. The information processing device 4 derives a correction value using the identified correction model 43 for prediction model 52a, and corrects the prediction values ​​from prediction model 52a using the derived correction value.

[0181] Figure 10 is a flowchart showing an example of the procedure for reconstructing the correction model 43. The processes in each flowchart below are executed by the processing unit 401 according to the program 44 stored in the storage unit 402 of the information processing device 4.

[0182] The processing unit 401 of the information processing device 4 receives from the operator's equipment 5 the identification information of the operator's equipment 5, operational data relating to one or more energy storage elements, the predicted value of the prediction model 52 owned by the operator of the operator's equipment 5, and the error between the predicted value and the measured value, in association with each other (step S111). Alternatively, the processing unit 401 may receive the identification information, operational data, the predicted value of the prediction model 52, and information for calculating the error between the predicted value and the measured value (e.g., the measured value).

[0183] The processing unit 401 reconstructs the correction model 43 based on the operational data acquired from the operator's equipment 5 and training data including the error between the predicted value and the measured value from the prediction model 52 (step S112). The processing unit 401 associates the reconstructed correction model 43 with the identification information of the operator's equipment 5, which is the source of the operational data, etc., and stores it in the storage unit 402 (step S113), and then terminates the process.

[0184] Figure 11 is a flowchart showing an example of the procedure for correcting the predicted values ​​of the prediction model 52. The information processing device 4 starts the following process when it receives a request for correction of the predicted values ​​from, for example, the operator's device 5.

[0185] The processing unit 401 of the information processing device 4 receives from the operator's equipment 5 the identification information of the operator's equipment 5, the operation data of the target energy storage element in the energy storage system 6, and the predicted value of the prediction model 52 owned by the operator of the operator's equipment 5, and associates them (step S121).

[0186] The processing unit 401 identifies a correction model 43 that corresponds to the identification information of the operator's equipment 5 from among a plurality of correction models 43 stored in the storage unit 402, based on the identification information of the operator's equipment 5 received (step S122). The processing unit 401 inputs the received operation data of the target energy storage element into the identified correction model 43 (step S123) and obtains the correction value output from the correction model 43 (step S124).

[0187] The processing unit 401 corrects the predicted values ​​of the prediction model 52 using the acquired correction values ​​(step S125). In step S125, the processing unit 401 corrects the predicted values ​​of the deterioration state by, for example, adding or subtracting the correction values ​​to the predicted values ​​of the deterioration state. The processing unit 401 transmits the prediction result, which represents the corrected predicted values, to the business operator's equipment 5, which is identified by the identification information of the business operator's equipment 5 (step S126). The processing unit 401 may, for example, generate a prediction result screen that includes the predicted values ​​of the prediction model 52 before correction, the predicted values ​​after correction, and the correction values, and provide the generated prediction result screen to the business operator's equipment 5.

[0188] The processing unit 401 calculates a usage fee based on the amount of predicted data from the prediction model 52 obtained in step S121, such that the fee increases with the amount of predicted data (step S127). The processing unit 401 associates the calculated usage fee with the identification information of the business operator's device 5 and outputs it to a predetermined output destination (step S128). The output destination for the usage fee may be, for example, a billing management system or the business operator's device 5. The usage fee may be determined according to the total amount of predicted data over a predetermined period. The processing unit 401 then completes the series of processes.

[0189] In the above, the prediction results are output to the operator's device 5. Alternatively, the output destination of the prediction results may be, for example, the display unit 404 of the information processing device 4, a user terminal used by operator X, etc. In other words, the prediction results may be provided to operator X.

[0190] (Modification 1) In Modification 1, when reconstructing the correction model 43, business operator X prepares the error between the predicted value and the measured value from the prediction model 52, or the measured value for determining the said error.

[0191] In the modified example 1, the information processing device 4 receives operational data and predicted values ​​from the prediction model 52 from the operator's device 5 when reconstructing the correction model 43. Operator X collects measured values ​​of the degradation state of the energy storage element by performing charging and discharging of the energy storage element in accordance with the operational data for a portion of the prediction target period for which predicted values ​​have been obtained. The information processing device 4 acquires the collected measured values, for example, by receiving operation input from the user of operator X.

[0192] The information processing device 4 calculates the error between the predicted values ​​of the prediction model 52 and the measured values ​​by comparing the acquired measured values ​​with the predicted values ​​of the prediction model 52 acquired from the operator's device 5, while synchronizing the time. Using the calculated error, the information processing device 4 updates the correction model 43 by executing the process shown in the flowchart of Figure 10.

[0193] With the above configuration, even if actual measurement values ​​are not available from businesses A to C, it is possible to collect error data and reconstruct the correction model 43. Based on the predicted values ​​of the prediction model 52 provided by businesses A to C and the error data provided by business X, the correction model 43 can be appropriately reconstructed.

[0194] This system is also applicable when business operator X uses energy storage elements from businesses A to C as reused items. The information processing device 4 obtains predicted values ​​of the prediction model 52 related to reused batteries from the business operator's device 5, obtains measured values ​​of the reused battery's condition after operation collected by business operator X, and updates the correction model by performing error analysis between these predicted and measured values. By correcting the predicted values ​​of the prediction model 52 using the updated correction model, the information processing device 4 can accurately obtain predicted values ​​that are in line with the actual operating conditions after reuse.

[0195] (Modification 2) In Modification 2, when reconstructing the correction model 43, operator X provides the predicted values ​​from the prediction model 42, and the error between the predicted values ​​and the measured values.

[0196] In the modified example 2, the information processing device 4 receives design information of the energy storage element used to generate the prediction model 52 from the operator's device 5 when reconstructing the correction model 43. The design information includes, for example, the configuration, materials, and applications of the energy storage element. The information processing device 4 obtains predicted values ​​of the degradation state by the prediction model 42 and measured values ​​for similar energy storage elements having a similar configuration, materials, and applications to the acquired design information. The predicted and measured values ​​for similar energy storage elements are obtained, for example, by extracting data related to energy storage elements with similar design information from test data already held by operator X.

[0197] The information processing device 4 calculates the error between the predicted value of the acquired prediction model 42 and the measured value, and uses the calculated error to update the correction model 43 by executing the flowchart in Figure 10, thereby reconstructing the correction model 43 for a specific prediction model 52.

[0198] According to the above configuration, even if both predicted values ​​and error data from the prediction model 42 cannot be obtained from businesses A to C, the correction model 43 can be updated by utilizing the prediction model 42 and test data held by business X.

[0199] Furthermore, the following variations are also included. The first business operator and the second business operator may or may not be the same. The following are examples of cases where the first business operator and the second business operator are the same. Even if a correction model has not been constructed in the prediction model being used, a new correction model can be constructed using the method of this disclosure. Alternatively, even if a correction model has already been constructed, if there is a problem with reproducibility (i.e., the error is large), it can be replaced with a better correction model using the method of this disclosure. For example, if the error of the correction model exceeds a certain level, the correction model may be automatically reconstructed using the method of this disclosure, or a system administrator or maintenance person may reconstruct the correction model using the method of this disclosure, taking the error into consideration.

[0200] (Fourth Embodiment) Figure 12 is a diagram showing an overview of the prediction system 300 of the fourth embodiment. The prediction system 300 of the fourth embodiment includes a prediction device 8 that predicts the lifespan of the energy storage devices 7 to be predicted. The number of energy storage devices 7 to be predicted is not limited.

[0201] The energy storage device 7 that the prediction system 300 is intended to predict comprises a rechargeable energy storage element 70 (cell), such as a secondary battery like a lead-acid battery or a lithium-ion battery. Alternatively, the energy storage element 70 may be a battery cell made of an all-solid-state battery, lead-acid battery, redox flow battery, zinc-air battery, lithium-sulfur battery, sodium-sulfur battery, silver-zinc oxide battery, or nickel-metal hydride battery, or it may be a capacitor. The energy storage device 7 may also be a group of energy storage elements comprising multiple energy storage elements 70, such as a module in which multiple cells are connected in series, a bank in which multiple modules are connected in series, or a domain in which multiple banks are connected in parallel.

[0202] An appropriate configuration can be adopted for the energy storage device 7. For example, if the energy storage device 7 is a module (i.e., a battery pack) in which multiple cells are connected in series, the module configuration may be a constrained type in which each cell is constrained in a state of compression in the thickness direction by a restraining member or the like, or an unconstrained type in which each cell is not constrained and is in an uncompressed state. The configuration of the energy storage device 7 is determined, for example, based on the application of the energy storage device 7.

[0203] Each of the energy storage devices 7 may be used for various purposes. Examples of applications for the energy storage devices 7 include power or auxiliary equipment for mobile vehicles such as automobiles, trains, and aircraft, industrial equipment such as construction machinery and forklifts, uninterruptible power supplies (UPS), thermal power generation systems, mega solar power generation systems, wind power generation systems, and other renewable energy power generation systems.

[0204] The energy storage device 7 includes measuring devices, a management device, and a communication device (not shown). The communication device may be integrated with the management device. The measuring device measures the current, voltage, and temperature of the energy storage device in a time series and outputs the measurement data, including the measured current, voltage, and temperature, to the management device. The management device can provide the acquired measurement data to the prediction device 8 via the communication device.

[0205] The prediction device 8 is an information processing device capable of various information processing and information transmission / reception related to the lifespan prediction of the energy storage device 7. The prediction device 8 is, for example, a server computer, a personal computer, a quantum computer, etc. The prediction device 8 is connected to the communication device of the energy storage device 7 via network N3. Network N3 is a wired or wireless network, including, for example, the internet, a carrier network that realizes wireless communication according to a predetermined mobile communication standard, or a general optical line. Network N3 may also include a local network for the manufacturer or maintenance provider of the energy storage device 7.

[0206] In this embodiment, the prediction device 8 is separate from the energy storage device 7. Alternatively, a management device provided on the energy storage device 7 may function as the prediction device 8. The management device that can function as the prediction device 8 may be, for example, a bank BMU (Battery Management Unit), a domain BMU, etc.

[0207] Figure 13 is a block diagram showing an example configuration of the prediction device 8. The prediction device 8 comprises a processing unit 81, a storage unit 82, a communication unit 83, a display unit 84, and an operation unit 85.

[0208] The processing unit 81 comprises one or more processors such as a CPU (Central Processing Unit) and a GPU (Graphics Processing Unit). The processing unit 81 includes memory, which is a temporary storage medium such as SRAM (Static Random Access Memory) and DRAM (Dynamic Random Access Memory). The processing unit 81 may also include functions such as a timer for measuring the elapsed time from the time a measurement start instruction is given to the time a measurement end instruction is given, a counter for counting numbers, and a clock for outputting date and time information. The processing unit 81 may be implemented in software, or part or all of it may be implemented in hardware such as an ASIC (Application Specific Integrated Circuit) or FPGA (Field Programmable Gate Array).

[0209] The storage unit 82 includes, for example, a non-volatile storage device such as a hard disk or flash memory. The storage unit 82 is separate from the prediction device 8 and may be one or more externally connected external storage devices. The storage unit 82 stores various computer programs and data that the processing unit 81 refers to. In this embodiment, the storage unit 82 stores a program 821, a correspondence table 822, and a prediction model 823 for causing a computer to execute processing related to predicting the lifespan of the energy storage device 7.

[0210] A computer program (program product) including program 821 may be provided on a non-temporary recording medium 8A on which the computer program is recorded in a readable format. The recording medium 8A is a portable memory such as a CD-ROM, USB memory, or SD (Secure Digital) card. The processing unit 81 reads the desired computer program from the recording medium 8A using a reading device (not shown) and stores the read computer program in the storage unit 82. Alternatively, the computer program may be provided by communication. Program 821 may be a single computer program or may consist of multiple computer programs. Program 821 may also be executed on a single computer or may be executed collaboratively by multiple computers.

[0211] The communication unit 83 includes a communication device that enables communication via the network N3. The processing unit 81 sends and receives data to and from the energy storage device 7 via the communication unit 83.

[0212] The display unit 84 includes a display device such as a liquid crystal display or an organic EL (Electro-Luminescence) display. The display unit 84 displays various information according to instructions from the processing unit 81.

[0213] The operation unit 85 is an interface that receives user input. The operation unit 85 includes, for example, a keyboard, mouse, touch panel device with a built-in display, speaker, and microphone. The operation unit 85 receives user input and sends control signals to the processing unit 81 according to the content of the operation.

[0214] The prediction device 8 may be configured to receive operations via an externally connected computer and output information to be notified to the external computer. In this case, the prediction device 8 does not need to have a display unit 84 and an operation unit 85.

[0215] The prediction model 823 is a model for predicting the state of the energy storage device 7. The prediction model 823 includes multiple individual prediction models 824, each corresponding to one of the multiple state predictions.

[0216] Each individual prediction model 824 is a model for performing a series of processes that output specific predicted values ​​using a specific theoretical formula. The predicted values ​​of the individual prediction model 824 may include numerical values, classification values, etc., corresponding to the prediction items.

[0217] Individual prediction models 824 are distinguished by the type of input data used for prediction and the type of prediction item to be output. For example, even if the same theoretical formula is used, if at least one of the contents of the input data and the contents of the output data are different, they are considered different individual prediction models 824.

[0218] In the example shown in Figure 13, the individual prediction model 824 includes a degradation prediction model 824a, a thickness prediction model 824b, a reaction force prediction model 824c, a first charge / discharge performance prediction model 824d, a second charge / discharge performance prediction model 824e, and a third charge / discharge performance prediction model 824f. The degradation prediction model 824a corresponds to the second prediction model, while the thickness prediction model 824b, the reaction force prediction model 824c, the first charge / discharge performance prediction model 824d, the second charge / discharge performance prediction model 824e, and the third charge / discharge performance prediction model 824f correspond to the first prediction model.

[0219] The degradation prediction model 824a is a model that predicts the time-dependent progression of the degradation state of the energy storage device 7. The predicted degradation value predicted by the degradation prediction model 824a is the value of the energy storage capacity of the energy storage device 7. Alternatively, the predicted degradation value may be State of Health (SOH), internal resistance, charge / discharge characteristics, amount of degradation of energy storage capacity (e.g., degradation amount when energized and degradation amount when not energized), degradation rate of positive and negative electrodes, deviation of capacity balance, etc. The degradation prediction model 824a can be used to predict various types of energy storage devices 7, regardless of their configuration and application.

[0220] The degradation prediction model 824a can be constructed using an appropriate method. For example, the technology described in Japanese Patent Publication No. 6428957 and Japanese Patent Publication No. 7173180 may be used as the degradation prediction model 824a. In the technology described in the above publications, the current, voltage, and temperature of the energy storage device 7 when charged and discharged according to the assumed power pattern are predicted by simulation based on the assumed power pattern flowing through the energy storage device 7 during the prediction period. Based on the SOC and temperature of the energy storage device 7 during the prediction period, which are predicted based on the obtained current, the progression of degradation of the energy storage device 7 is predicted.

[0221] The thickness prediction model 824b is a model that predicts the change in the thickness of the energy storage device 7 over time. The thickness prediction model 824b can be constructed using an appropriate method. For example, the thickness prediction model 824b may use the technology described in Japanese Patent No. 7392363. In the technology described in the above publication, time-series data of the SOC in the energy storage device 7 is acquired, the fluctuation range of the SOC in the acquired time-series data and a representative value of the SOC that represents the SOC region in the fluctuation range are identified, and the thickness of the energy storage device 7 is estimated based on the identified fluctuation range and representative value.

[0222] The reaction force prediction model 824c is a model that predicts the temporal changes in the reaction force of the energy storage device 7. The reaction force is the pressure generated in a constrained energy storage device 7 due to the expansion and contraction of the energy storage elements accompanying charging and discharging. The reaction force prediction model 824c can be constructed using an appropriate method. The reaction force prediction model 824c may, for example, use the technology described in Japanese Patent No. 7392364. In the technology described in the above publication, time-series data of the SOC in the energy storage device 7 is acquired, the fluctuation range of the SOC in the acquired time-series data and a representative value of the SOC that represents the SOC region in the fluctuation range are identified, and the reaction force of the energy storage device 7 is estimated based on the identified fluctuation range and representative value.

[0223] The first charge / discharge performance prediction model 824d, the second charge / discharge performance prediction model 824e, and the third charge / discharge performance prediction model 824f are models that predict the time-dependent changes in the charge / discharge performance of the energy storage device 7, using a model current pattern or a model load as input. Charge / discharge performance refers to, for example, the charge acceptance capacity or discharge capacity of the energy storage device 7 at a certain point in the prediction period.

[0224] The charge and discharge performance can be represented by various predicted values ​​depending on the application of the energy storage device 7, based on various input elements depending on the application of the energy storage device 7. Multiple charge and discharge performance prediction models are provided to correspond to each prediction. The charge and discharge performance includes, for example, current value, voltage value, internal resistance value, classification value of whether power can be supplied or not, etc. Specifically, the charge and discharge performance includes current value, voltage value, internal resistance value, or classification value of whether power can be supplied or not when a model current pattern is applied or a model load is applied.

[0225] Each of the first charge / discharge performance prediction models 824d to the third charge / discharge performance prediction model 824f can be constructed using an appropriate method. Each of the first charge / discharge performance prediction models 824d to the third charge / discharge performance prediction model 824f may use, for example, the technology described in Japanese Patent No. 7568039. In the technology described in the above publication, time-series data of the SOC in the energy storage device 7 is acquired, the fluctuation range of the SOC in the acquired time-series data and a representative value of the SOC that represents the SOC region in the fluctuation range are identified, and the charge / discharge performance of the energy storage element is estimated based on the identified fluctuation range and representative value. In this embodiment, the first charge / discharge performance prediction model 824d, the second charge / discharge performance prediction model 824e, and the third charge / discharge performance prediction model 824f each predict different types of charge / discharge performance.

[0226] In the prediction method of this embodiment, the lifespan of the energy storage device 7 is predicted based on two or more indicators, including a predicted degradation value by the degradation prediction model 824a and a predicted state value by an individual prediction model 824 other than the degradation prediction model 824a. Hereinafter, the individual prediction model 824 other than the degradation prediction model 824a will also be referred to as the other individual prediction model 824. The type of the other individual prediction model 824 is determined according to the configuration and application of the energy storage device 7. The other individual prediction model 824 used in combination with the degradation prediction model 824a includes one or more other individual prediction models 824 selected according to the configuration of the energy storage device 7 and one or more other individual prediction models 824 selected according to the application of the energy storage device 7.

[0227] The prediction device 8 has pre-stored a correspondence table 822 in the storage unit 82 that records the correspondence between at least one of the configuration and application of the energy storage device 7 and an individual prediction model 824 used to predict the lifespan of the energy storage device 7 for that configuration and application. By referring to the correspondence table 822, the prediction device 8 can identify the individual prediction model 824 to be used to predict the lifespan of the energy storage device 7 to be predicted, according to the type of configuration and application of the energy storage device 7.

[0228] Figure 14 shows an example of the information stored in the correspondence table 822. The correspondence table 822 stores the configuration or application of the energy storage device 7 in association with the individual prediction model 824. In the example in Figure 14, the correspondence between the configuration and the prediction model is as follows: the unconstrained type is associated with the thickness prediction model 824b, and the constrained type is associated with the reaction force prediction model 824c. Furthermore, the correspondence between the application and the prediction model is as follows: the vehicle is associated with the first charge / discharge performance prediction model 824d, the vehicle auxiliary equipment is associated with the second charge / discharge performance prediction model 824e, and the UPS is associated with the third charge / discharge performance prediction model 824f. The degradation prediction model 824a is associated with all configurations and applications.

[0229] In this embodiment, two individual prediction models 824, corresponding to the configuration and application of the energy storage device 7, are used as other individual prediction models 824 used together with the degradation prediction model 824a. Alternatively, the number of individual prediction models 824 used in combination with the degradation prediction model 824a may be one or three or more. For example, one individual prediction model 824 corresponding to the configuration or application of the energy storage device 7 may be used, or two individual prediction models 824 corresponding to the configuration of the energy storage device 7 may be used.

[0230] Figure 15 is a flowchart showing an example of a processing procedure performed by the prediction device 8. The processing in each flowchart below is performed by the processing unit 81 according to the program 821 stored in the storage unit 82 of the prediction device 8. The prediction device 8 starts the following processing in response to receiving a request for life prediction, for example, when a user operates the operation unit 85.

[0231] The processing unit 81 of the prediction device 8 acquires the configuration and application corresponding to the energy storage device 7 to be predicted by receiving input from the user, for example via the operation unit 85 (step S211). The processing unit 81 may also receive the configuration and application by communication with an external device. For example, the processing unit 81 acquires either a constrained or unconstrained configuration for the energy storage device 7, and acquires either a vehicle, vehicle auxiliary equipment, or UPS for the application of the energy storage device 7.

[0232] The processing unit 81 identifies the type of other individual prediction model 824 that corresponds to the constraint type or unconstrained type of configuration of the acquired energy storage device 7, based on the information stored in the correspondence table 822 (step S212). In step S212, either the thickness prediction model 824b corresponding to the unconstrained type or the reaction force prediction model 824c corresponding to the constraint type is identified.

[0233] The processing unit 81 identifies the type of other individual prediction model 824 that corresponds to the vehicle, vehicle accessory, or UPS, which is the application of the acquired energy storage device 7, based on the information stored in the correspondence table 822 (step S213). In step S213, one of the following is identified: the first charge / discharge performance prediction model 824d corresponding to a vehicle, the second charge / discharge performance prediction model 824e corresponding to a vehicle accessory, or the third charge / discharge performance prediction model 824f corresponding to a UPS.

[0234] The processing unit 81 selects from among a plurality of individual prediction models 824 stored in the storage unit 82 another individual prediction model 824 that corresponds to the type of other individual prediction model 824 that was identified, namely the thickness prediction model 824b or the reaction force prediction model 824c (step S214). The processing unit 81 uses the selected thickness prediction model 824b or reaction force prediction model 824c to predict the thickness or reaction force of the energy storage device 7 (step S215).

[0235] The processing unit 81 selects from among a plurality of individual prediction models 824 stored in the storage unit 82 another individual prediction model 824 corresponding to the type of other individual prediction model 824 identified, namely the first charge / discharge performance prediction model 824d, the second charge / discharge performance prediction model 824e, or the third charge / discharge performance prediction model 824f (step S216). The processing unit 81 uses the selected first charge / discharge performance prediction model 824d, the second charge / discharge performance prediction model 824e, or the third charge / discharge performance prediction model 824f to predict the charge / discharge performance of the energy storage device 7 (step S217). In step S217, the processing unit 81 may, for example, obtain a constantly changing required performance given by a higher-level device and predict the charge acceptance capacity or discharge capacity according to the obtained required performance.

[0236] The processing unit 81 selects a degradation prediction model 824a from among a plurality of individual prediction models 824 stored in the storage unit 82 (step S218). The processing unit 81 uses the selected degradation prediction model 824a to predict the capacity degradation of the energy storage device 7 (step S219).

[0237] The processing unit 81 predicts the lifespan of the energy storage device 7 based on the predicted thickness or reaction force of the energy storage device 7, the charge / discharge performance, and the capacity degradation (step S220). In step S220, the processing unit 81 identifies an operational area that satisfies all of the following conditions: for example, the predicted value of the thickness or reaction force is within a preset threshold range, the predicted value of the charge / discharge performance is within a preset threshold range, and the predicted value of the capacity degradation is within a preset threshold range. The processing unit 81 finally determines the lifespan of the energy storage device 7 based on the duration of the identified operational area. By identifying an area where each predicted value satisfies predetermined operating conditions, an operational area that comprehensively evaluates all predictive indicators can be determined. The conditions for each predicted value used to identify the operational area can be set appropriately depending on the type of predicted value. The conditions for each predicted value may include conditions such as being greater than or equal to a threshold, being less than a threshold, or being a specific classification value.

[0238] The processing unit 81 displays the lifespan prediction result on the display unit 84 (step S221) and completes the series of processes. The output destination for the lifespan prediction result is not limited to the display unit 84, but may be an external device such as a terminal device used by the user.

[0239] In the life prediction process of step S220 described above, the processing unit 81 may further consider the predicted temperature of the energy storage device 7. The temperature of the energy storage device 7 may include the temperature of the energy storage device 7 or the area surrounding the energy storage device 7, the amount of change in said temperature (e.g., temperature rise), etc. The processing unit 81 may identify a region that satisfies all of the following conditions: the predicted temperature of the energy storage device 7 is within a preset threshold range, the predicted thickness or reaction force is within a preset threshold range, the predicted charge / discharge performance is within a preset threshold range, and the predicted capacity degradation is within a preset threshold range.

[0240] Other individual prediction models 824 may be configured to receive predicted degradation values ​​from degradation prediction model 824a and output predicted state values ​​corresponding to the received predicted degradation values. The processing unit 81 can input the predicted degradation values ​​to the other individual prediction models 824, thereby reflecting the capacity degradation prediction results in the prediction of state values.

[0241] Similarly, the degradation prediction model 824a may be configured to accept predicted state values ​​from individual prediction models 824 other than the degradation prediction model 824a, and to output predicted degradation values ​​corresponding to the accepted predicted state values. By inputting the predicted state values ​​to the degradation prediction model 824a, the processing unit 81 can reflect the prediction results of other state values ​​in the prediction of capacity degradation.

[0242] By linking the degradation prediction model 824a with other individual prediction models 824, the final prediction accuracy can be improved compared to calculating the predicted degradation value and predicted state value independently.

[0243] The main entity performing each process in the flowchart above is not limited to the prediction device 8; some or all of the above processes may be performed by, for example, the management device for the energy storage device 7. The prediction device 8 and the management device for the energy storage device 7 may cooperate to perform a series of processes.

[0244] For example, the management device for the energy storage device 7 stores each degradation prediction model 824a to the third charge / discharge performance prediction model 824f in advance. The prediction device 8 identifies the types of other individual prediction models 824 through the processing from steps S211 to S213 and transmits information indicating the types of other individual prediction models 824 identified to the management device for the energy storage device 7. When the management device for the energy storage device 7 receives information indicating the types of other individual prediction models 824 from the prediction device 8, it performs various predictions using the degradation prediction models 824a to the third charge / discharge performance prediction model 824f that it has stored in advance by executing the processing from step S215 onward.

[0245] The embodiments disclosed herein should be considered in all respects to be illustrative and not restrictive. The technical features described in each embodiment can be combined with each other, and the scope of the present invention is intended to include all modifications within the claims and equivalents thereof. The sequences shown in each embodiment are not limiting, and within a consistent scope, each processing step may be performed in a different order, and multiple processes may be performed in parallel. The processing entities for each process are not limiting, and within a consistent scope, the processing of each device may be performed by other devices.

[0246] The matters described in each embodiment can be combined with each other. Furthermore, the independent and dependent claims described in the claims can be combined with each other in any combination, regardless of the form of reference. In addition, the claims use a form in which claims referencing two or more other claims (multi-claim form), but are not limited to this. A form in which multi-claims referencing at least one multi-claim (multi-multi-claim) may also be used.

[0247] 100 Prediction System 1 Energy Storage Device 10 Energy Storage Element 2 Prediction Device 222 Prediction Model 223 Correction Model 200 Energy Storage Information Processing System 4 Information Processing Device 42 Prediction Model (First Prediction Model) 43 Correction Model 5a-5c Operator Equipment 52a-52c Prediction Model (Second Prediction Model) 6a-6c,6x Energy Storage System 300 Prediction System 7 Energy Storage Device 70 Energy Storage Element 8 Prediction Device 822 Correspondence Table 823 Prediction Model 824 Individual Prediction Model

Claims

1. A prediction device comprising a processing unit that acquires a first predicted value using a prediction model for predicting the state of an energy storage element, estimates a first correction value for correcting the error of the acquired first predicted value using a correction model for correcting the error of the prediction value of the prediction model, and corrects the first predicted value using the estimated first correction value.

2. The prediction device according to claim 1, wherein the correction model is a model that outputs a correction value according to the usage history data of the energy storage element.

3. The prediction device according to claim 1 or 2, wherein the prediction model is updated when the error between the corrected first predicted value and the measured value of the state of the energy storage element is outside the acceptable range.

4. The prediction device according to claim 3, wherein when the prediction model is updated, the correction model is updated based on the predicted values ​​of the updated prediction model.

5. The prediction device according to claim 1 or 2, which determines whether the first error between the corrected first predicted value and the measured value of the state of the energy storage element is within an acceptable range, and if it is determined that the first error is outside the acceptable range, updates the parameters in the prediction model, which is shown by a calculation formula including predetermined parameters, and determines whether the second error between the second predicted value from the prediction model after updating the parameters and the measured value of the state of the energy storage element is within an acceptable range, and if it is determined that the second error is outside the acceptable range, updates the calculation formula in the prediction model.

6. The prediction device according to claim 5, wherein if it is determined that the second error is outside the acceptable range, it estimates a second correction value for correcting the error of the second predicted value using the correction model, corrects the second predicted value by the prediction model after updating the parameters using the estimated second correction value, determines whether the third error between the corrected second predicted value and the measured value of the state of the energy storage element is within the acceptable range, and updates the calculation formula in the prediction model if it is determined that the third error is outside the acceptable range.

7. The prediction device according to claim 1 or 2, which outputs information indicating an abnormal state of the energy storage element when the update frequency of the correction model or the prediction model, or the predicted value obtained using the updated correction model or the prediction model, matches the abnormal state.

8. A prediction method comprising: obtaining a first predicted value using a prediction model that predicts the state of an energy storage element; estimating a first correction value for correcting the error of the obtained first predicted value using a correction model that corrects the error of the prediction value of the prediction model; and correcting the first predicted value using the estimated first correction value.

9. A computer program that causes a computer to perform the following processes: obtain a first predicted value using a prediction model that predicts the state of an energy storage element; estimate a first correction value to correct the error of the obtained first predicted value using a correction model that corrects the error of the prediction value of the prediction model; and correct the first predicted value using the estimated first correction value.

10. An information processing method comprising: an information processing device of a first business operator acquiring predicted values ​​predicted by a second prediction model of a second business operator for predicting the state of an energy storage element; generating correction information for the acquired predicted values ​​using a correction model constructed based on the error between the predicted values ​​predicted by a first prediction model of a first business operator for predicting the state of an energy storage element and the measured value of the state of the energy storage element; and outputting the generated correction information.

11. The information processing method according to claim 10, which involves obtaining the error between the predicted value predicted by the second prediction model and the measured value of the state of the energy storage element, reconstructing the correction model based on the obtained error between the predicted value of the second prediction model and the measured value, and generating the correction information using the reconstructed correction model.

12. The information processing method according to claim 10, wherein the correction model is a model that takes operational data indicating the operational status of the energy storage element as input and outputs the error between the predicted value of a prediction model that predicts the state of the energy storage element and the measured value.

13. The information processing method according to claim 11, which obtains a predicted value predicted by the second prediction model provided by the second business operator and obtains the error provided by the first business operator.

14. An information processing method according to claim 10 or 11, comprising: acquiring design information of an energy storage element used in constructing the second prediction model; acquiring the error between the predicted value predicted by the first prediction model and the measured value for the state of another energy storage element having design information similar to the acquired design information; reconstructing the correction model based on the error between the predicted value and the measured value of the acquired first prediction model; and generating the correction information using the reconstructed correction model.

15. The information processing method according to claim 10 or claim 11, which outputs the correction information for the predicted values ​​obtained from the second business operator to the second business operator, and calculates the usage fee such that the fee becomes higher the larger the amount of data of the predicted values ​​obtained from the second business operator.

16. A computer program that causes a computer to acquire predicted values ​​predicted by a second prediction model of a second business operator for predicting the state of an energy storage element, generate correction information for the acquired predicted values ​​using a correction model generated based on the error between the predicted values ​​predicted by a first prediction model of a first business operator for predicting the state of an energy storage element and the measured values ​​of the state of the energy storage element, and output the generated correction information.

17. An information processing device comprising a processing unit that acquires predicted values ​​predicted by a second prediction model of a second business operator for predicting the state of an energy storage element, generates correction information for the acquired predicted values ​​using a correction model generated based on the error between the predicted values ​​predicted by a first prediction model of a first business operator for predicting the state of an energy storage element and the measured value of the state of the energy storage element, and outputs the generated correction information.

18. A prediction method for predicting the lifespan of an energy storage device, which involves obtaining the structure or application of an energy storage device, identifying a first prediction model corresponding to the type of structure or application obtained from among a plurality of first prediction models for predicting conditions other than degradation in the energy storage device, and predicting the lifespan of the energy storage device based on the predicted state value of the energy storage device predicted by the identified first prediction model and the predicted degradation value of the energy storage device predicted by a second prediction model for predicting the degradation of the energy storage device.

19. The prediction method according to claim 18, wherein the first prediction model includes a model for predicting the thickness of the energy storage device and a model for predicting the reaction force of the energy storage device, and the model for predicting the thickness or the model for predicting the reaction force is selected according to the type of structure of the energy storage device.

20. The prediction method according to claim 19, wherein when an unconstrained type is obtained as the structure of the energy storage device, a model for predicting the thickness is selected, and when a constrained type is obtained as the structure of the energy storage device, a model for predicting the reaction force is selected.

21. The prediction method according to claim 18 or 19, wherein the first prediction model includes a model for predicting the charge and discharge performance of different types of energy storage devices, and the first prediction model is selected from a plurality of models for predicting charge and discharge performance according to the type of application of the energy storage device.

22. The prediction method according to claim 18 or 19, wherein the predicted state value of the energy storage device includes the thickness or reaction force of the energy storage device and the charge / discharge performance.

23. The prediction method according to claim 18 or 19, wherein the lifespan of the energy storage device is predicted such that the predicted state value of the energy storage device satisfies predetermined requirements and the predicted degradation value of the energy storage device satisfies predetermined conditions.

24. The prediction method according to claim 18 or 19, which involves obtaining a predicted temperature value of the energy storage device and predicting the lifespan of the energy storage device such that the obtained predicted temperature value of the energy storage device, the predicted state value of the energy storage device, and the predicted degradation value of the energy storage device each satisfy predetermined conditions.

25. A prediction method according to claim 18 or 19, wherein a plurality of the first prediction models are stored in advance, a first prediction model is selected from the plurality of stored first prediction models, and the predicted state value of the energy storage device is predicted using the selected first prediction model.

26. A prediction device comprising a processing unit that acquires the structure or application of an energy storage device, identifies a first prediction model corresponding to the acquired type of structure or application from among a plurality of first prediction models for predicting conditions other than degradation in the energy storage device, and performs a process to predict the lifespan of the energy storage device based on the predicted state value of the energy storage device predicted by the identified first prediction model and the predicted degradation value of the energy storage device predicted by a second prediction model for predicting the degradation of the energy storage device.

27. A computer program that obtains the structure or application of an energy storage device, identifies a first prediction model corresponding to the type of structure or application obtained from among a plurality of first prediction models for predicting conditions other than degradation in the energy storage device, and causes a computer to perform a process of predicting the lifespan of the energy storage device based on the predicted state value of the energy storage device predicted by the identified first prediction model and the predicted degradation value of the energy storage device predicted by a second prediction model for predicting the degradation of the energy storage device.

Images