Battery self-discharge prediction method and related equipment

Abstract

The present disclosure provides a battery self-discharge prediction method and related equipment, which relate to the field of energy storage technology. The method includes obtaining a variety of process data of a battery to be tested during the production process, the process data including process parameter data of at least one of the film production stage, the assembly stage and the formation stage; extracting features from the various process data to obtain a plurality of process features, and calculating the feature importance value of each process feature; based on the feature importance value of each process feature, screening target process features whose feature importance values ​​meet preset conditions, and screening target process data corresponding to the target process features; processing the target process data through a self-discharge prediction model to obtain a self-discharge prediction value of the battery to be tested, and the self-discharge prediction model is obtained by training a variety of historical process data of battery samples and corresponding battery sample self-discharge labels.

Description

Background Art

[0002] Battery aging is a natural phenomenon during use. Battery self-discharge refers to the phenomenon that the battery capacity decreases spontaneously after the battery cell is charged in an open circuit state and left idle for a period of time. It represents the charge retention ability of the battery cell under certain conditions. Batteries with high self-discharge rates usually lead to rapid capacity decay, which can cause energy storage system failure.

[0003] In related technologies, the capacity retention method, open circuit voltage decay method, current measurement method, etc. are usually used to detect self-discharge of batteries at the factory stage and in laboratory scenarios. However, the above detection methods have problems such as long measurement time, high equipment sensitivity requirements, and high production costs due to energy and equipment occupation. Summary of the invention

[0004] The present disclosure provides a battery self-discharge prediction method and related equipment, which at least to a certain extent overcome the problems of long test time, high equipment sensitivity requirement and high production cost in the battery self-discharge detection method in the related art.

[0005] Other features and advantages of the present disclosure will become apparent from the following detailed description, or may be learned in part by the practice of the present disclosure.

[0006] According to one aspect of the present disclosure, a battery self-discharge prediction method is provided, comprising: obtaining a variety of process data of a battery to be tested during a production process, the process data comprising process parameter data of at least one of a film production stage, an assembly stage, and a formation stage; performing feature extraction on the various process data to obtain a plurality of process features; calculating a feature importance value of each of the process features to battery self-discharge; screening target process features whose feature importance values ​​meet preset conditions, and screening target process data corresponding to the target process features; processing the target process data through a self-discharge prediction model to obtain a self-discharge prediction value of the battery to be tested, wherein the self-discharge prediction model is trained by a variety of historical process data of battery samples and corresponding battery sample self-discharge labels.

[0007] In one embodiment of the present disclosure, the self-discharge prediction model is trained in the following manner: determining multiple initial hyperparameter combinations of the self-discharge prediction model, the initial hyperparameter combinations being multiple data groups consisting of different values ​​of multiple hyperparameters of the self-discharge prediction model; constructing a data set based on multiple historical process data and corresponding battery sample self-discharge labels, the data set including a training data set and a verification data set; traversing the multiple initial hyperparameter combinations, iteratively training the self-discharge prediction model to be trained based on the training data set until convergence, and obtaining the self-discharge prediction model corresponding to each of the initial hyperparameter combinations; processing the verification data set through the self-discharge prediction models corresponding to each of the initial hyperparameter combinations, and obtaining the evaluation index score of the self-discharge prediction model corresponding to each of the initial hyperparameter combinations; taking the initial hyperparameter combination with the highest evaluation index score as the optimal hyperparameter combination, and taking the self-discharge prediction model corresponding to the optimal hyperparameter combination as the trained self-discharge prediction model.

[0008] In one embodiment of the present disclosure, the hyperparameters of the self-discharge prediction model include at least one of the learning rate, the maximum depth of the decision tree, the number of decision trees, the minimum sample weight and of the leaf nodes, the sample sampling ratio of each decision tree, the feature sampling ratio of each decision tree, the minimum splitting loss, and the weight of the regularization term.

[0009] In one embodiment of the present disclosure, the evaluation index of the self-discharge prediction model includes at least one of mean square error, determination coefficient, accuracy, and recall rate.

[0010] In one embodiment of the present disclosure, the method further includes: regularly updating the data set based on multiple process data of the battery to be tested and the corresponding battery self-discharge labels, and retraining the self-discharge prediction model using the updated data set.

[0011] In one embodiment of the present disclosure, the target process features whose feature importance values ​​meet preset conditions based on the feature importance values ​​of each process feature to battery self-discharge are screened, including any one of the following methods: if the feature importance value is greater than or equal to a preset feature importance value threshold, then the feature importance value is determined to meet the preset condition; based on the feature importance value, a plurality of process features are sorted in descending order, and if the sorting order of the feature importance value is less than or equal to a first preset order threshold, then the feature importance value is determined to meet the preset condition; based on the feature importance value, a plurality of process features are sorted in ascending order, and if the sorting order of the feature importance value is greater than or equal to the preset order threshold, then the feature importance value is determined to meet the preset condition.

[0012] In one embodiment of the present disclosure, the self-discharge prediction model is an extreme gradient boosting model; wherein, the calculation of the feature importance value of each of the process features to the battery self-discharge includes any one of the following methods: counting the number of times each of the process features is used at a split node in all decision trees, and taking the number of times used as the feature importance value; calculating the information gain of each of the process features, and taking the information gain as the feature importance value, wherein the information gain represents the average reduction in the loss function value of each of the process features when it is used as a split node; calculating the coverage of each of the process features, and taking the coverage as the feature importance value, wherein the coverage represents the average number of samples covered when each of the process features is used as a split node in all decision trees.

[0013] In one embodiment of the present disclosure, the process characteristics include raw material ratio, stirring time, coating thickness, rolling density, cutting accuracy, battery cell size, electrode coating uniformity, pole piece alignment, winding tension, welding quality, sealing process, formation conditions, capacity separation test results, hot melt pressure, hot melt time, total baking time, preheating time, cathode water capacity, dielectric withstand voltage test time, electrolyte water content, electrolyte hydrofluoric acid, preset temperature standing time, formation starting voltage, formation capacity, end temperature, full welding power, capacity after load adjustment, capacity starting voltage, discharge capacity, charging capacity, standing time, open circuit voltage, open circuit voltage test temperature, battery cell temperature, helium inspection production time, vacuum baking production time, preset temperature standing production time, formation production time, capacity production time, room temperature standing production time, and open circuit voltage or insulation pulse test generation time. One or more of the time.

[0014] In one embodiment of the present disclosure, after the target process data is processed by the self-discharge prediction model to obtain the self-discharge prediction value of the battery to be tested, the method further includes: visually displaying the self-discharge prediction value and the self-discharge label of the battery to be tested, so that the user can make production decisions based on the self-discharge prediction value and the self-discharge label of the battery to be tested.

[0015] According to another aspect of the present disclosure, a battery self-discharge prediction device is provided, including: a data acquisition module, used to acquire a variety of process data of a battery to be tested during the production process, the process data including process parameter data of at least one of a film production stage, an assembly stage and a formation stage; a feature extraction module, used to perform feature extraction on the various process data to obtain a plurality of process features, and calculate the feature importance value of each of the process features; a data screening module, used to screen target process features whose feature importance values ​​meet preset conditions, and screen target process data corresponding to the target process features; a model prediction module, used to process the target process data through a self-discharge prediction model to obtain a self-discharge prediction value of the battery to be tested, wherein the self-discharge prediction model is obtained by training a variety of historical process data of battery samples with corresponding battery sample self-discharge labels.

[0016] According to another aspect of the present disclosure, an electronic device is also provided, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to execute any one of the above-mentioned battery self-discharge prediction methods by executing the executable instructions.

[0017] According to another aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the battery self-discharge prediction method described in any one of the above is implemented.

[0018] According to another aspect of the present disclosure, a computer program product is provided, including a computer program, wherein when the computer program is executed by a processor, the battery self-discharge prediction method described above is implemented.

[0019] In the embodiment of the present disclosure, a variety of process data of the battery to be tested in the production process are obtained, and the process data include process parameter data of at least one of the film production stage, the assembly stage and the formation stage; feature extraction is performed on the various process data to obtain multiple process features, and the feature importance value of each process feature is calculated; the target process feature whose feature importance value meets the preset conditions is screened, and the target process data corresponding to the target process feature is screened; the target process data is processed by the self-discharge prediction model to obtain the self-discharge prediction value of the battery to be tested, and the self-discharge prediction model is obtained by training the various historical process data of the battery sample and the corresponding battery sample self-discharge label. In the present disclosure, on the one hand, the process data in the battery production process is processed by the self-discharge prediction model, which effectively improves the accuracy and prediction efficiency of the self-discharge prediction. On the other hand, by calculating the feature importance value of the process feature and screening the target process feature according to the feature importance value, effective process parameter information can be extracted, thereby screening the process feature that has a greater impact on the battery self-discharge, and further improving the accuracy of the battery self-discharge prediction.

[0020] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The accompanying drawings herein are incorporated into the specification and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the specification are used to explain the principles of the present disclosure. Obviously, the accompanying drawings described below are only some embodiments of the present disclosure, and for ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without creative work.

[0022] Figure 1 A schematic structural diagram of an energy storage system provided by an embodiment of the present disclosure is shown.

[0023] Figure 2 A flow chart of a battery self-discharge prediction method provided by an embodiment of the present disclosure is shown.

[0024] Figure 3 A flow chart of another battery self-discharge prediction method provided by an embodiment of the present disclosure is shown.

[0025] Figure 4 A flow chart of another battery self-discharge prediction method provided by an embodiment of the present disclosure is shown.

[0026] Figure 5 A flow chart of another battery self-discharge prediction method provided by an embodiment of the present disclosure is shown.

[0027] Figure 6 A flow chart of a self-discharge prediction model training method provided in an embodiment of the present disclosure is shown.

[0028] Figure 7 A flow chart of another battery self-discharge prediction method provided in an embodiment of the present disclosure is shown.

[0029] Figure 8 An example flow chart of a battery self-discharge prediction method provided in an embodiment of the present disclosure is shown.

[0030] Fig. 9 A schematic diagram showing the prediction effect of a self-discharge prediction model provided in an embodiment of the present disclosure is shown.

[0031] Fig.10 A graph showing the relationship between the predicted K value and the actual K value provided in the embodiments of the present disclosure is shown.

[0032] Fig.11 A graph showing the relationship between the error size and the number of samples provided in an embodiment of the present disclosure is shown.

[0033] Fig.12 The distribution density curves of the actual K value and the predicted K value provided in the embodiments of the present disclosure are shown.

[0034] Fig.13 A schematic structural diagram of a battery self-discharge prediction device provided by an embodiment of the present disclosure is shown.

[0035] Fig.14 A structural block diagram of an electronic device provided by an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0036] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that the disclosure will be more comprehensive and complete and to fully convey the concepts of the example embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0037] In addition, the accompanying drawings are only schematic illustrations of the present disclosure and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor devices and / or microcontroller devices.

[0038] Since the energy people need is highly temporal and spatial, in order to rationally use energy and improve energy utilization, it is necessary to use a medium or device to store one form of energy in the same form or convert it into another form of energy, and then release it in a specific form based on future application needs. At present, the main way to generate green electricity is to develop green energy such as photovoltaics and wind power to replace fossil energy.

[0039] At present, the generation of green electricity generally relies on photovoltaics, wind power, and hydropower, etc. However, wind and solar energy generally have problems of strong intermittency and large volatility, which will cause instability in the power grid, insufficient electricity during peak hours, and too much electricity during valley hours. Unstable voltage will also cause damage to electricity. Therefore, insufficient electricity demand or insufficient grid acceptance capacity may cause the problem of "abandoning wind and solar power". To solve these problems, we must rely on energy storage. That is, converting electrical energy into other forms of energy through physical or chemical means and storing it, and converting energy into electrical energy and releasing it when needed. In simple terms, energy storage is like a large "power bank". When photovoltaic and wind energy are sufficient, electrical energy is stored and the stored electricity is released when needed.

[0040] Taking electrochemical energy storage as an example, the present disclosure provides an energy storage device, which has a group of chemical batteries inside. The chemical elements in the batteries are mainly used as energy storage media. The charging and discharging process is accompanied by chemical reactions or changes in the energy storage media. In simple terms, the electricity generated by wind and solar energy is stored in chemical batteries. When the use of external electricity reaches a peak, the stored electricity is released for use, or transferred to places where electricity is scarce for use.

[0041] At present, energy storage (i.e. energy storage) has a wide range of application scenarios, including power generation side energy storage, grid side energy storage and power consumption side energy storage. The corresponding types of energy storage equipment include:

[0042] (1) Large-scale energy storage power stations applied to wind power and photovoltaic power stations can help renewable energy generation meet grid connection requirements and improve the utilization rate of renewable energy. As a high-quality active / reactive regulating power source on the power supply side, energy storage power stations can achieve load matching of electric energy in time and space, enhance the ability to absorb renewable energy, reduce instantaneous power changes, reduce the impact on the power grid, improve the absorption of new energy generation, and are of great significance in power grid system backup, relieving peak load power supply pressure, and peak and frequency regulation.

[0043] (2) Energy storage containers used on the grid side mainly perform peak load regulation, frequency regulation, and grid congestion relief. They can achieve peak load shaving and valley filling of power loads, that is, charging the energy storage battery when the power load is low and releasing the stored electricity during the peak load period, thereby achieving a balance between power production and consumption, such as energy storage power station systems;

[0044] (3) Small energy storage cabinets used on the power consumption side, mainly for self-generation and self-use of electricity, peak-valley price arbitrage, capacity cost management and improving power supply reliability. According to different application scenarios, energy storage on the power consumption side can be divided into industrial and commercial energy storage cabinets, household energy storage equipment, energy storage charging piles, etc., which are generally used in conjunction with distributed photovoltaics. Industrial and commercial users can use energy storage for valley-peak price arbitrage and capacity cost management. In the power market that implements peak-valley electricity prices, by charging the energy storage system at low electricity prices and discharging the energy storage system at high electricity prices, peak-valley electricity price arbitrage is achieved to reduce electricity costs. In addition, industrial enterprises that are subject to two-part electricity prices can use energy storage systems to store energy during low electricity consumption and discharge at peak loads, thereby reducing peak power and the maximum demand reported, and achieving the purpose of reducing capacity electricity charges. Household photovoltaic storage can improve the level of self-generation and self-use of electricity. Due to high electricity prices and poor power supply stability, the demand for household photovoltaic installations has been driven. Considering that photovoltaic power is generated during the day, and users generally have higher loads at night, energy storage can be configured to better utilize photovoltaic power, improve the level of self-generation and self-use, and reduce electricity costs. In addition, communication base stations, data centers and other fields need to configure energy storage for backup power.

[0045] See also Figure 1 , Figure 1 is a schematic diagram of the structure of an energy storage system according to an embodiment of the present disclosure, and Figure 1 The embodiment is described by taking the energy storage scenario in user-side energy storage as an example. The energy storage system disclosed in the present invention is not limited to its household energy storage scenario, but can also be applied to industrial and commercial energy storage scenarios.

[0046] The present disclosure provides an energy storage system, which includes: an energy storage device 1, an electric energy conversion device 2 (such as a photovoltaic panel), a first user load 3 (such as a street lamp), a second user load 4 (such as an air conditioner and other household appliances), etc. The energy storage device 1 is a small energy storage box that can be mounted on an outdoor wall by wall hanging. Specifically, the photovoltaic panel can convert solar energy into electric energy during the period of low electricity prices, and the energy storage device 1 is used to store the electric energy and supply it to street lamps and household appliances for use during the peak electricity prices, or to supply power when the power grid is off / power outage.

[0047] The number of energy storage devices 1 can be multiple, and multiple energy storage devices 1 are connected in series or in parallel. Multiple energy storage devices 1 are supported and electrically connected by isolation plates (not shown). In this embodiment, "multiple" refers to two or more. An energy storage box can also be provided outside the energy storage device 1 to accommodate the energy storage device 1.

[0048] Optionally, the energy storage device 1 may include but is not limited to a single cell, a battery module, a battery pack, a battery system, etc. The actual application form of the energy storage device 1 provided in the embodiment of the present disclosure may be but is not limited to the listed products, and may also be other application forms. The embodiment of the present disclosure does not strictly limit the application form of the energy storage device 1. The embodiment of the present disclosure is only described by taking the energy storage device 1 as a battery cell as an example. When the energy storage device 1 is a single cell, the energy storage device 1 may be at least one of a cylindrical battery, a square battery, etc.

[0049] During the battery production process, multiple tests need to be performed after the battery is formed, i.e., after the first charge and activation, such as self-discharge test, capacity test, and internal resistance test. The self-discharge test is also called the K value test, and the internal resistance test is also called the DC impedance test or the DC internal resistance (DCR) test. Through the above quality control measures, batteries with possible defects can be screened out, effectively preventing defective batteries from entering the market and causing safety accidents.

[0050] The self-discharge of a battery refers to the phenomenon that the battery loses power due to internal chemical reactions when the battery is not connected to an external circuit. It is also called charge retention capacity. The self-discharge reaction of a battery is irreversible, which not only leads to a reduction in the capacity of the battery itself, but also seriously affects the battery assembly and cycle life. Therefore, during the battery production process, it is necessary to detect the self-discharge value of the battery.

[0051] In related technologies, the capacity retention method, open circuit voltage decay method, current measurement method, etc. are usually used for battery self-discharge detection before the battery leaves the factory. For example, when the open circuit voltage decay method is used for K value testing, the processing flow includes:

[0052] Aging treatment: Newly produced batteries go through a period of charge and discharge cycles, a process called aging. The purpose of aging treatment is to make the internal chemical reaction of the battery more stable and the electrode material fully activated;

[0053] Standing: After the aging treatment is completed, the battery needs to be left standing for a period of time under predetermined conditions, such as a few hours to a day, so that the internal state of the battery can be stabilized and the electrochemical balance can be restored;

[0054] Open circuit voltage measurement: After the battery is at rest, the K value of the battery is obtained by measuring the open circuit voltage of the battery (i.e. the voltage value of the battery when it is not connected to an external circuit). The K value is the ratio of the battery open circuit voltage to the standard voltage curve, reflecting the internal resistance and capacity characteristics of the battery;

[0055] Data analysis: In order to calculate the K value of the battery, the measured open circuit voltage value needs to be compared with the standard characteristic curve of the battery. The K value calculation formula is: K = (OCV2-OCV1) / t, where t represents the time interval of the battery being stationary, OCV1 is the open circuit voltage value before the battery is stationary, and OCV2 is the open circuit voltage value after the battery is stationary. It should be noted that OCV1 and OCV2 are both standard battery voltage values ​​measured under the same conditions. In this way, the K value reflecting the battery performance can be accurately calculated;

[0056] Result recording: After calculating the K value, record the result on the battery label or in the database for subsequent battery classification, grouping and quality control.

[0057] However, for the open circuit voltage decay method, the aging treatment and battery rest process require a long time, which will increase the generation cycle, reduce the generation efficiency, and consume a lot of time; the long aging treatment and rest process will increase energy consumption and equipment occupancy, and the production cost is high; due to individual differences in batteries, even after the same aging and rest process, the consistency of the battery may still be poor; in the process of open circuit voltage measurement and data recording, the data may be inaccurate due to operator errors.

[0058] In order to solve at least part of the above technical problems, the battery self-discharge prediction method provided by the embodiment of the present disclosure obtains a variety of process data of the battery to be tested in the production process, and the process data includes process parameter data of at least one of the film production stage, assembly stage and formation stage; extracts features of the various process data to obtain multiple process features, and calculates the feature importance value of each process feature; selects the target process feature whose feature importance value meets the preset conditions, and selects the target process data corresponding to the target process feature; processes the target process data through the self-discharge prediction model to obtain the self-discharge prediction value of the battery to be tested, and the self-discharge prediction model is obtained by training the various historical process data of the battery sample and the corresponding battery sample self-discharge label. In the present disclosure, on the one hand, the process data in the battery production process is processed by the self-discharge prediction model, which effectively improves the accuracy and prediction efficiency of the self-discharge prediction. On the other hand, by calculating the feature importance value of the process feature and selecting the target process feature according to the feature importance value, effective process parameter information can be extracted, thereby selecting the process feature that has a greater impact on the battery self-discharge, and further improving the accuracy of the battery self-discharge prediction.

[0059] In the present disclosure, it can be applied to the prediction of the self-discharge value of lithium iron phosphate battery cells, and can also be applied to the prediction of the self-discharge value of nickel-cadmium batteries, nickel-hydrogen batteries, etc., and the present disclosure does not make specific limitations on this.

[0060] Under the above system architecture, a battery self-discharge prediction method is provided in the embodiment of the present disclosure, and the method can be executed by any electronic device with computing and processing capabilities, such as a processor, etc. The processor can be a programmable logic controller (PLC) or a single-chip microcomputer, etc., as long as it can realize the corresponding functions.

[0061] Figure 2 A flow chart of a method for predicting battery self-discharge in an embodiment of the present disclosure is shown as follows: Figure 2 As shown, the battery self-discharge prediction method provided in the embodiment of the present disclosure includes the following steps:

[0062] S202, obtaining a variety of process data of the battery to be tested during the production process, wherein the process data includes process parameter data of at least one of the wafer manufacturing stage, the assembly stage and the formation stage.

[0063] The battery to be tested is a battery for which self-discharge prediction is to be performed.

[0064] The lithium battery production process mainly includes the following stages: film making stage, assembly stage, formation stage and packaging stage. The film making stage includes positive electrode batching, positive electrode coating, positive electrode film making, positive electrode sheet preparation, negative electrode batching, negative electrode coating, negative electrode film making, and negative electrode sheet preparation; the assembly stage includes winding, shell entry, spot welding, baking, liquid injection, cap welding, and sealing; the formation stage includes primary cleaning, shell coding, dry storage, secondary cleaning, heat shrinkage, pre-charging, OCV0, first high temperature storage, second normal temperature storage, coding, data loading, OCV1, third normal temperature storage, OCV2, fourth normal temperature storage, OCV3, capacity division, OCV4 and grading; the packaging process includes grade code spraying, grade scanning inspection, packaging and warehousing.

[0065] In one embodiment, the various process data are actual process parameter data of process parameters that need to be controlled in battery production in the above-mentioned wafer production stage, assembly stage, and formation stage.

[0066] For example, in the positive electrode preparation and negative electrode preparation processes, the process data may be process parameter data such as raw material ratio and stirring time; in the positive electrode coating and negative electrode coating processes, the process data may include process parameter data such as coating thickness, rolling density, and cutting accuracy; in the formation stage, the process data may include process parameter data such as formation conditions and formation capacity.

[0067] It should be noted that process data can be collected through the upstream manufacturing execution system to obtain detailed process parameter data during the battery production process.

[0068] In one embodiment, the processor may integrate the process data collected by the manufacturing execution system, and the integrated process data may be stored in the form of a data table or a key-value pair. The integrated data includes a battery identifier and a plurality of process data corresponding to the battery identifier.

[0069] S204, extracting features from a variety of process data to obtain a plurality of process features, and calculating a feature importance value of each process feature.

[0070] The various process data are actual process parameter data of process parameters that need to be controlled in battery production in the above-mentioned film production stage, assembly stage and formation stage. Correspondingly, the process characteristics that are controlled during the battery production process may include raw material ratio, stirring time, coating thickness, rolling density, cutting accuracy, battery cell size, electrode coating uniformity, pole piece alignment, winding tension, welding quality, sealing process, formation conditions, capacity separation test results, hot melt pressure, hot melt time, total baking time, preheating time, cathode water capacity, dielectric withstand voltage test time, electrolyte water content, electrolyte hydrofluoric acid, preset temperature standing time, formation starting voltage, One or more of the formation capacity, end temperature, full welding power, capacity after load adjustment, capacity start voltage, discharge capacity, charge capacity, standing time, open circuit voltage, open circuit voltage test temperature, battery cell temperature, helium inspection production time, vacuum baking production time, preset temperature standing production time, formation production time, capacity production time, normal temperature standing production time, and open circuit voltage or insulation pulse test generation time, by setting process data of multiple processes in multiple production stages, it is possible to screen process features that have a large impact on the battery self-discharge value, thereby effectively improving the accuracy of battery self-discharge prediction. It should be noted that in the formation stage, according to process requirements, the battery to be tested can be placed in a high temperature environment of a preset temperature for multiple times. For example, the above-mentioned preset temperature standing time can include a first preset temperature standing time and a second preset temperature standing time. The first preset temperature standing time and the second preset temperature standing time can be standing times in different high temperature environments, such as 65°C, 85°C, etc. Normal temperature standing means that the battery needs to be left standing for a certain period of time after filling or formation to ensure that the electrolyte can fully infiltrate the battery. The normal temperature standing production time can be the normal temperature standing time after the battery is filled, or the normal temperature standing time after the battery is formed.

[0071] The sealing of the battery can be tested by helium testing technology. In battery production, two helium tests are usually required. One helium test is performed before the battery is filled with liquid to test the sealing of the welds around the battery cover; the final helium test is performed after the battery is filled with liquid to test the sealing of the filling port. The helium test production time refers to the time consumed when performing helium testing on the battery, which can include the time consumed for the first helium test production and the final helium test production.

[0072] The dielectric withstand voltage test, also known as the Hi-pot test, is a key quality inspection process for bare cells after subsequent assembly. It can detect whether there are foreign particles inside the battery, whether the diaphragm is damaged, or whether there are burrs on the edges of several pieces, and other factors that may cause internal short circuits in the battery, thereby ensuring the safety of the battery during use.

[0073] The open circuit voltage OCV test is a test of the open circuit voltage, AC internal resistance, and shell voltage of the battery cell. It needs to meet the open circuit voltage accuracy of 0.1mV and the shell voltage accuracy of 1MV. Among them, the open circuit voltage OCV1 refers to the open circuit voltage value measured at time t1, and the open circuit voltage OCV2 refers to the open circuit voltage value measured at time t2.

[0074] It should be noted that the above process characteristics can be determined according to actual circumstances and are not specifically limited in this disclosure.

[0075] In one embodiment, after the integrated process data is obtained, the data may be preprocessed, and the preprocessing methods include data cleaning, filling missing values, processing outliers, data standardization, feature engineering, etc.

[0076] The purpose of data cleaning is to improve the quality and reliability of the data set, ensure the accuracy and effectiveness of data analysis, mainly remove irrelevant data, eliminate redundant information that is not related to the prediction of battery self-discharge value, and reduce the complexity of data processing.

[0077] Filling missing values ​​is used to process missing values ​​in the integrated process data, remove duplicate data, and avoid deviations in battery self-discharge values.

[0078] When dealing with outliers, outliers may be due to measurement error, data entry errors, or other reasons.

[0079] Data standardization is used to standardize or normalize the integrated process data so that the data is on the same scale for easy comparison and analysis. Standardization is the ratio of the difference between the feature and the mean and the standard deviation. The StandardScaler class of the preprocessing library can be used to standardize the data.

[0080] The purpose of feature engineering is to extract the most important information for the self-discharge prediction model from the integrated process data and convert it into a form suitable for processing by machine learning algorithms. The integrated process data is then preliminarily explored to understand the basic situation of the data, such as distribution, trends, and outliers.

[0081] Feature engineering also includes feature selection, feature construction and feature extraction. Feature selection can use filtering, packaging, embedding and other methods to screen features with high correlation with battery self-discharge values, and select features with strong correlation with battery self-discharge values ​​from multiple process features. The filtering method scores each feature according to divergence or correlation, sets a threshold or the number of selection thresholds, and selects the target process feature. The packaging method selects a number of features or excludes a number of features each time according to the objective function, and then obtains the target process feature. The embedding method uses a machine learning algorithm to train the model, obtains the weight coefficients of each feature, and selects the target process feature from large to small according to the weight coefficients. In addition, statistical tests and other methods can also be used for feature selection.

[0082] In one embodiment, in feature engineering, the feature importance value of each process feature is calculated, and feature selection is performed based on the feature importance value. The feature importance value is used to characterize the importance of the process feature to the battery self-discharge value. The larger the feature importance value, the higher the influence of the process feature on the battery self-discharge value, and the smaller the feature importance value, the lower the influence of the process feature on the battery self-discharge value.

[0083] The process data is combined through feature engineering to extract effective process parameter information as an important characteristic factor affecting the battery self-discharge value. The characteristic factor varies with the change of process data.

[0084] S206 , screening target process features whose feature importance values ​​meet preset conditions, and screening target process data corresponding to the target process features.

[0085] In one embodiment, the above-mentioned preset conditions can be determined according to actual needs and can be pre-configured in the processor so that the processor can quickly perform feature selection.

[0086] The target process characteristic is a process characteristic whose characteristic importance value meets preset conditions, and the target process data is process data corresponding to the target process characteristic; in other words, the target process characteristic is a process characteristic that has a greater degree of influence on the battery self-discharge value.

[0087] S208, processing the target process data through a self-discharge prediction model to obtain a self-discharge prediction value of the battery to be tested, wherein the self-discharge prediction model is trained by various historical process data of battery samples and corresponding battery sample self-discharge labels.

[0088] In one embodiment, the self-discharge prediction model takes various historical process data as input and battery self-discharge values ​​as output, and represents the corresponding relationship between the various historical process data and the battery self-discharge values.

[0089] The self-discharge prediction model can select a machine learning model according to actual needs. For example, the self-discharge prediction model can be an extreme gradient boosting model (eXtreme Gradient Boosting, XGBoost). XGBoost is a gradient boosting decision tree (Gradient Boosting Decision Tree, GBDT) based learner. It continuously adds new decision trees to fit the residual of the previous model, thereby achieving model optimization. The decision tree of the XGBoost model adopts the Level-wise growth method. In each iteration, it will try to split all leaf nodes.

[0090] The extreme gradient boosting model uses a regression tree for each base classifier, and the training method uses the forward step-by-step algorithm to optimize each base classifier.

[0091] For the i-th sample in the t-th tree, the prediction value of the XGBoost model is expressed as:

[0092]

[0093] in, is the prediction result of sample i after the tth iteration, f t (x i ) is the model prediction result of the t-th decision tree, is the prediction result of the t-1th decision tree.

[0094] For the i-th sample, the final prediction value is:

[0095]

[0096] Among them, T is the number of decision trees

[0097] The objective function Obj of the XGBoost model is:

[0098]

[0099] Where n is the total number of samples. The first term in formula 3 is the loss function. The loss function can be a mean square error function or other types of loss functions. The second term is a regularization term, which represents the complexity of all t decision trees. The loss function part is for all samples, and the regularization part is for all decision trees.

[0100] In one embodiment, the self-discharge prediction model can also use a decision-enhanced CatBoost model, a lightweight gradient boosting machine LightGBM, etc. The model training and model prediction methods are the same as the XGBoost model type, which will not be repeated here.

[0101] In the embodiment of the present disclosure, a variety of process data of the battery to be tested in the production process are obtained, and the process data include process parameter data of at least one of the film production stage, the assembly stage and the formation stage; feature extraction is performed on the various process data to obtain multiple process features, and the feature importance value of each process feature is calculated; the target process feature whose feature importance value meets the preset conditions is screened, and the target process data corresponding to the target process feature is screened; the target process data is processed by a self-discharge prediction model to obtain the self-discharge prediction value of the battery to be tested, and the self-discharge prediction model is obtained by training the various historical process data of the battery sample and the corresponding battery sample self-discharge label. On the one hand, the present disclosure processes the process data in the battery production process by the self-discharge prediction model, effectively improving the accuracy and prediction efficiency of the self-discharge prediction. On the other hand, by calculating the feature importance value of the process feature and screening the target process feature according to the feature importance value, effective process parameter information can be extracted, thereby screening the process feature that has a greater impact on the battery self-discharge, and further improving the accuracy of the battery self-discharge prediction.

[0102] Figure 3 FIG. 2 is a flow chart of another method for predicting battery self-discharge provided by an embodiment of the present disclosure. Figure 3 As shown, in one embodiment, the self-discharge prediction model is an extreme gradient boosting model; wherein the above S204 calculates the feature importance value of each process feature, including:

[0103] S2042. Count the number of times each process feature is used at a split node in all decision trees, and use the number of times used as the feature importance value.

[0104] In all decision trees, each process feature can be split as a partition attribute, and the number of times it is used can be expressed as Weight. For example, in 350 decision trees, the number of times the electrolyte water content is used as a split node is 200, and the number of times the full welding power is used as a split node is 280. 200 and 280 are the feature importance values ​​of the electrolyte water content and the full welding power, respectively. The feature importance value of the full welding power is greater than that of the electrolyte water content, indicating that the influence of the full welding power on the battery self-discharge value is greater than that of the electrolyte water content, and the probability of full welding power screening as the target process feature is higher.

[0105] In the embodiment of the present disclosure, by counting the number of times each process feature in all decision trees is used at the splitting node, the feature importance value of each process feature can be quickly determined, thereby improving the battery self-discharge prediction efficiency.

[0106] Figure 4 FIG. 2 is a flow chart of another method for predicting battery self-discharge provided by an embodiment of the present disclosure. Figure 4As shown, in one embodiment, the above S204 calculates the feature importance value of each process feature, including:

[0107] S2044. Calculate the information gain of each process feature and use the information gain as the feature importance value. The information gain represents the average reduction in the loss function value of each process feature when it is used as a split node.

[0108] In one embodiment, information gain is also called Gain. Taking process feature k=1, 2, ...K as an example, information gain can be calculated by the following formula:

[0109]

[0110] Where k represents the kth node, T represents the number of all decision trees, N(t) represents the number of non-leaf nodes in the tth decision tree, β(t,i) represents the partitioning feature of the ith non-leaf node in the tth decision tree, β(·)∈1,2,…,K, K is a positive integer, I(·) is the indicator function, G γ(t,i) H represents the sum of the first-order derivatives of all samples falling on the i-th non-leaf node of the t-th decision tree, γ(t,i) G represents the sum of the second-order derivatives of all samples falling on the i-th non-leaf node of the t-th decision tree. γ(t,i,L) represents the sum of the first-order derivatives of the left nodes of the i-th non-leaf node of the t-th decision tree, G γ(t,i,r) represents the sum of the first-order derivatives of the right nodes of the i-th non-leaf node of the t-th decision tree, H γ(t,i,L) represents the sum of the second-order derivatives of the left node of the i-th non-leaf node of the t-th decision tree, H γ(t,i,R) represents the sum of the second-order derivatives falling on the right node of the i-th non-leaf node of the t-th decision tree, and λ is the hyperparameter of the regularization term.

[0111] If the information gain of the process feature is higher, it means that the process feature has a higher degree of influence on the battery self-discharge value. If the information gain of the process feature is smaller, it means that the feature has a lower degree of influence on the battery self-discharge value. Therefore, based on the information gain as the feature importance and as the condition for screening the target process features, the process features with a high degree of influence on the battery self-discharge value are obtained, thereby improving the accuracy of battery self-discharge prediction.

[0112] Figure 5 FIG. 2 is a flow chart of another method for predicting battery self-discharge provided by an embodiment of the present disclosure. Figure 5 As shown, in one embodiment, the above S204 calculates the feature importance value of each process feature, including:

[0113] S2046. Calculate the coverage of each process feature and use the coverage as the feature importance value. The coverage represents the average number of samples covered when each process feature is used as a split node in all decision trees.

[0114] In one embodiment, coverage is also called Cover. When the decision tree splits, the number of samples covered by the leaf nodes under the process feature is divided by the number of times the process feature is used to split. The closer the split is to the root, the greater the coverage. The greater the coverage, the greater the feature importance value, and the greater the influence of the process feature on the battery self-discharge value.

[0115] Coverage can be defined as the average of the sum of the second-order derivatives of the process feature corresponding to the sample when it is used as a partitioning attribute. Coverage can be calculated by the following formula:

[0116]

[0117] Where k represents the kth node, T represents the number of all decision trees, N(t) represents the number of non-leaf nodes in the tth decision tree, β(t,i) represents the partitioning feature of the ith non-leaf node in the tth decision tree, β(·)∈1,2,…,K, K is a positive integer, I(·) is the indicator function, H γ(r,i) Represents the sum of the second-order derivatives of all samples falling on the i-th non-leaf node of the t-th decision tree.

[0118] In the embodiments of the present disclosure, coverage is used as a feature importance value to effectively evaluate the importance of process features in all decision trees. Based on coverage as a condition for screening target process features, process features with a high degree of influence on battery self-discharge values ​​are obtained, thereby improving the accuracy of battery self-discharge prediction.

[0119] In one embodiment, the screening of target process features whose feature importance values ​​meet preset conditions in S206 includes any of the following methods:

[0120] If the feature importance value is greater than or equal to the preset feature importance value threshold, it is determined that the feature importance value meets the preset condition;

[0121] Sort the plurality of process features in descending order based on the feature importance values, and if the sorting order of the feature importance values ​​is less than or equal to a first preset order threshold, then determine that the feature importance values ​​meet a preset condition;

[0122] The plurality of process features are sorted in ascending order based on the feature importance values. If the sorting order of the feature importance values ​​is greater than or equal to a second preset order threshold, it is determined that the feature importance values ​​meet a preset condition.

[0123] The above-mentioned preset feature importance value threshold can be pre-configured in the processor. The preset feature importance value threshold can be determined according to actual needs, and the present disclosure does not make specific limitations on this. For example, the preset feature importance value threshold is 0.2.

[0124] After the feature importance values ​​of multiple process features are sorted in descending order, the sorting order of the multiple feature importance values ​​is arranged from small to large. As the sorting order increases, the feature importance value gradually decreases. At this time, the process feature with a preset order threshold can be selected as the target process feature. For example, the process feature with the feature importance value ranked in the top 5 (top5) is selected as the target process feature.

[0125] After the feature importance values ​​of multiple process features are sorted in ascending order, the sorting order of the multiple feature importance values ​​is arranged from small to large. As the sorting order increases, the feature importance value also gradually increases. At this time, the process feature corresponding to the feature importance value whose sorting order is greater than or equal to the second preset order threshold is used as the target process feature. For example, the last five process features in the sorting order are selected as the target process features.

[0126] It should be noted that the above-mentioned first preset order threshold and second preset order threshold can be configured in the processor and determined according to actual needs, and the present disclosure does not make specific limitations on this.

[0127] In the embodiments of the present disclosure, by screening process features whose feature importance values ​​meet preset conditions in a variety of ways, the efficiency of feature selection can be improved, and process features with a high degree of influence on battery self-discharge can be effectively determined, thereby improving the accuracy of battery self-discharge prediction.

[0128] Figure 6 A flow chart of a self-discharge prediction model training method provided in an embodiment of the present disclosure is shown. Figure 6 As shown, in one embodiment, the self-discharge prediction model is trained in the following manner:

[0129] S602, determining a plurality of initial hyperparameter combinations of the self-discharge prediction model, where the initial hyperparameter combinations are a plurality of data groups formed by different values ​​of a plurality of hyperparameters of the self-discharge prediction model;

[0130] S604, constructing a data set based on a variety of historical process data and corresponding battery sample self-discharge labels, the data set including a training data set and a verification data set;

[0131] S606, traversing multiple initial hyperparameter combinations, iteratively training the self-discharge prediction model to be trained based on the training data set until convergence, and obtaining the self-discharge prediction model corresponding to each initial hyperparameter combination;

[0132] S608, processing the validation data set by using the self-discharge prediction model corresponding to each initial hyperparameter combination to obtain an evaluation index score of the self-discharge prediction model corresponding to each initial hyperparameter combination;

[0133] S610: Taking the initial hyperparameter combination with the highest evaluation index score as the optimal hyperparameter combination, and taking the self-discharge prediction model corresponding to the optimal hyperparameter combination as the trained self-discharge prediction model.

[0134] In one embodiment, in S602, the self-discharge prediction model may include multiple hyperparameters. Based on different values ​​of each hyperparameter, multiple hyperparameters are combined to obtain multiple initial hyperparameter combinations. The training process of the self-discharge prediction model also includes the process of selecting the optimal hyperparameter combination from the multiple initial hyperparameter combinations.

[0135] In one embodiment, the hyperparameters of the self-discharge prediction model may include at least one of the learning rate learning_rate, the maximum depth of the decision tree max_depth, the number of decision trees, the minimum sample weight and of the leaf nodes, the sample sampling ratio of each decision tree, the feature sampling ratio of each decision tree, the minimum split loss, and the weight of the regularization term, so that by selecting the hyperparameters that have a greater impact on the self-discharge prediction model, a model that is more suitable for battery self-discharge prediction can be trained to improve the prediction accuracy of the self-discharge prediction model.

[0136] Among them, the learning rate learning_rate controls the contribution of each decision tree to the self-discharge prediction model, and takes a floating point number between 0 and 1. When learning_rate is small, such as 0.01 to 0.1, the robustness of the model can be improved; the maximum depth max_depth of the decision tree affects the complexity of the decision tree, and can take a positive integer between 3 and 20; the number of decision trees, that is, the number of base learners, can take a positive integer between 50 and 1000; the minimum sample weight of the leaf node is used to prevent overfitting, and can be a positive integer between 1 and 10; the sample sampling ratio of each decision tree is a floating point number between 0.5 and 1, which can increase the robustness of the model; the feature sampling ratio of each decision tree is a floating point number between 0.5 and 1, which helps to prevent overfitting; the minimum split loss controls the growth of the decision tree, and the decision tree will only split when the loss function value drops by more than the minimum split loss, and can take a non-negative number between 0 and 5.

[0137] The weight of the regularization term may include at least one of the weight of the L1 regularization term and the weight of the L2 regularization term. The weight of the L1 regularization term is used for sparse process feature selection and can be a non-negative number between 0 and 5; the weight of the L2 regularization term is used to control the complexity of the model and can be a non-negative number between 0 and 5.

[0138] For example, if the hyperparameters of the self-discharge prediction model include learning rate, maximum depth of decision tree, and number of decision trees, the hyperparameter combination can be represented by an array form {learning_rate, max_depth, T}. When the above three hyperparameters take different values, multiple arrays are formed, which are called initial hyperparameter combinations.

[0139] In S604, a data set is constructed based on a plurality of historical process data and corresponding battery sample self-discharge labels. The data set includes a training data set and a verification data set. The number of samples in the training data set and the verification data set can be determined according to a preset ratio. For example, the samples in the training data set and the verification data set are determined according to a ratio of 8:2 or 7:3. The training data set is used to train the self-discharge prediction model, and the verification data set is used to verify the prediction effect of the self-discharge prediction model.

[0140] The battery sample self-discharge label is the true value of the self-discharge value measured by the corresponding battery sample.

[0141] In S606, one of the initial hyperparameter combinations can be randomly selected from the multiple initial hyperparameter combinations, and a self-discharge prediction model to be trained can be constructed based on the initial hyperparameter combination, and the self-discharge prediction model to be trained can be iteratively trained based on the training data set, with the goal of minimizing the prediction error, until the model converges or reaches a preset stop condition, and the self-discharge prediction model corresponding to the initial hyperparameter combination is obtained. Multiple initial hyperparameter combinations are traversed, and the self-discharge prediction model to be trained corresponding to each initial hyperparameter combination is iteratively trained based on the training data set until convergence, and the self-discharge prediction model corresponding to each initial hyperparameter combination is obtained.

[0142] Model convergence means that the prediction error of the self-discharge prediction model no longer decreases. The preset stop condition can be that the iterative training duration reaches a preset duration threshold, or the number of iterative training times reaches a preset number threshold. The preset duration threshold or the preset number threshold can be determined according to actual needs, and the present disclosure does not make specific limitations on this.

[0143] During the self-discharge prediction model training process, the model parameters are continuously adjusted to minimize the prediction error. The model parameters may include weights or bias terms.

[0144] In S608, the validation data set is processed by the self-discharge prediction model corresponding to each initial hyperparameter combination to obtain the evaluation index score of the self-discharge prediction model corresponding to each initial hyperparameter combination. Overfitting and underfitting are avoided by cross-validation. Cross-validation is a common method for evaluating performance. By dividing the data set into a training data set and a validation data set, the performance of the self-discharge prediction model on unknown data is evaluated through the validation data set to ensure that the model can be effectively applied to actual scenarios. During the model evaluation process, the test set data is used to evaluate the actual performance of the model. The test set is a data set independent of the training set and is used to evaluate the generalization ability of the model. Through the evaluation of the test set, the performance of the model in actual applications can be understood, and corresponding decisions can be made accordingly.

[0145] It should be noted that the evaluation indicators of the self-discharge prediction model include mean square error MSE, determination coefficient R 2 , accuracy, and recall rate, and evaluate the prediction effect of the self-discharge prediction model through multiple evaluation indicators to improve the generalization ability of the model.

[0146] Mean square error (MSE) is an indicator that measures the difference between the predicted value and the true value (or label). The lower the MSE value, the more accurate the model's prediction. 2 It measures the ability of the model to explain the variability of the dependent variable. The closer it is to 1, the better the model's explanatory power. Accuracy and recall can be used for classification problems. Accuracy reflects the proportion of positive classes correctly predicted by the model, and recall reflects the proportion of correct predictions made by the model among all positive classes.

[0147] In S610, after obtaining the evaluation index scores of the self-discharge prediction models corresponding to the initial hyperparameter combinations, the initial hyperparameter combination with the highest evaluation index score is used as the optimal hyperparameter combination, and the self-discharge prediction model corresponding to the optimal hyperparameter combination is used as the trained self-discharge prediction model.

[0148] In the disclosed embodiment, during the self-discharge prediction model training process, hyperparameter search is performed through a grid parameter adjustment method to determine the optimal hyperparameter combination, so that the model can learn the complex patterns of the data, thereby providing more accurate prediction results and improving prediction accuracy.

[0149] In one embodiment, the method further includes: regularly updating the data set based on multiple process data of the battery to be tested and the corresponding battery self-discharge labels, and retraining the self-discharge prediction model using the updated data set.

[0150] As new data is continuously generated, the data set can be updated every month or quarter, and the self-discharge prediction model can be retrained using the new data set to reflect new trends and patterns in the data, ensuring the long-term effectiveness and performance improvement of the model, and continuously and effectively serving business needs. Model maintenance and optimization will help the model better adapt to data changes and improve the performance of the model in practical applications.

[0151] Figure 7 FIG. 2 is a flow chart of another method for predicting battery self-discharge provided in an embodiment of the present disclosure. Figure 7 As shown, in one embodiment, after the target process data is processed by the self-discharge prediction model in S208 to obtain the self-discharge prediction value of the battery to be tested, the method further includes:

[0152] S210 , visually displaying the self-discharge prediction value and the self-discharge label of the battery to be tested, so that the user can make production decisions according to the self-discharge prediction value and the self-discharge label of the battery to be tested.

[0153] In one embodiment, the battery self-discharge prediction value and the self-discharge label can be displayed on the display screen. The results can be displayed in the form of numbers and charts for easy understanding and comparison.

[0154] The battery's self-discharge label can also be displayed, and the predicted self-discharge value can be compared with the self-discharge label to ensure the accuracy and reliability of the prediction. Based on the verification results, production managers and quality control teams can make production decisions, such as further checking battery performance when there is a large difference between the predicted value and the actual value. Production parameters can also be adjusted based on the prediction results and verification conditions to continuously optimize the consistency and reliability of battery performance.

[0155] By continuously monitoring the performance of the model in actual operations and regularly evaluating the model's prediction accuracy on an independent validation dataset, we can ensure that the model's performance does not degrade over time. When analyzing errors in model predictions, it helps to understand the nature of the problem and take appropriate measures to make adjustments. Based on the results of performance monitoring and error analysis, we can adjust the model, for example, we can try different algorithms, feature selection, and feature engineering methods to improve the performance of the model, thereby continuously optimizing the model and improving its performance in actual applications.

[0156] In order to deepen the understanding of the battery self-discharge prediction method disclosed in the present invention, a detailed description is given below with reference to specific examples.

[0157] like Figure 8 As shown, the battery self-discharge prediction method disclosed in the present invention comprises the following steps:

[0158] S801. Acquire multiple historical process data of battery samples and corresponding battery sample self-discharge labels to construct a data set; the data set includes a training data set and a verification data set, the training data set and the verification data set divide the samples according to a preset ratio, and the historical process data includes historical process parameter data of at least one of the production stage, assembly stage and formation stage.

[0159] The data set may be stored in a data table format and may include a battery sample ID, various historical process data, and a battery sample self-discharge label.

[0160] Before constructing the data set, it also includes the steps of extracting features from multiple historical process data to obtain multiple historical process features, calculating feature importance values ​​of the historical process features, and screening target historical process features and corresponding historical feature data whose feature importance values ​​meet preset conditions, thereby constructing the data set.

[0161] Historical process characteristics include hot melt pressure, hot melt time, total baking time, preheating time, cathode water content 1, cathode water content 2, Hi-pot test time, electrolyte water content, electrolyte hydrofluoric acid, high-temperature standing 1 standing time, formation starting voltage, formation capacity, end temperature, high-temperature standing 2 time, full welding power, capacity after load adjustment, capacity starting voltage, ninth step discharge capacity, seventh step discharge capacity, fourth step charging capacity, fifth step charging capacity, fourth step capacity, standing time, OCV1, OCV1-temperature, battery cell temperature, first helium inspection production time, vacuum baking production time, high-temperature standing 1 production time, formation production time, high-temperature standing 2 production time, final helium inspection production time, capacity production time, normal temperature standing 1 production time, OCV1 / IMP1 test production time.

[0162] The feature importance values ​​are calculated, and the top 5 historical process features with the most important feature values ​​are screened out as shown in Table 1.

[0163] Table 1 Relationship between target process characteristics and characteristic importance values

[0164]

[0165] S802: Train the self-discharge prediction model based on the training data set to obtain a trained self-discharge prediction model.

[0166] The self-discharge prediction model may be trained with reference to S602 to S610 in the above embodiment.

[0167] You can set multiple initial hyperparameter combinations, perform hyperparameter search through grid adjustment, and determine the optimal hyperparameter combination.

[0168] For example, an XGBoost model can be defined, where the maximum depth of the decision tree max_depth = 13, the learning rate learning_rate = 0.05, and the number of decision trees n_estimators = 350.

[0169] After the optimal hyperparameter combination is determined, the self-discharge prediction model corresponding to the optimal hyperparameter combination can be used as the self-discharge prediction model finally trained.

[0170] S803, obtaining a plurality of process data of the battery to be tested, the process data also includes process parameter data of the film production stage, the assembly stage and the formation stage, which will not be described in detail here.

[0171] S804: predicting the self-discharge value of the battery to be tested by using the self-discharge prediction model to obtain a self-discharge prediction value.

[0172] The self-discharge value of the battery to be tested is predicted by using the patterns and rules learned by the self-discharge prediction model to obtain the self-discharge prediction value of the battery to be tested.

[0173] S805. Output the results and display them in the form of numbers or charts.

[0174] like Fig. 9 As shown, the result output includes the self-discharge prediction value (predicted k value) and the self-discharge label (real k value) of the battery to be tested.

[0175] S806: The self-discharge prediction value and self-discharge label of the battery to be tested are verified.

[0176] Compare the predicted value with the true value to verify the accuracy and reliability of the prediction result.

[0177] like Figure 10-12 As shown in the figure, by fitting the predicted k value and the true k value, we can get: the overall average accuracy MAPE is 97.79%; the minimum deviation rate is 0.00%; the maximum deviation rate is 22.31%; the overall average prediction error MAE is 0.000665; the cumulative absolute error < 0.001 accounts for 77.88%; the true k value < predicted k value accounts for 50.03%; the maximum k value difference is 0.006260; the minimum k value difference is -0.005351, and the determination coefficient R 2 It is 86.31%.

[0178] S807: Make a corresponding decision according to the predicted self-discharge value of the battery to be tested.

[0179] Based on the predicted k value, production managers and quality control teams can make corresponding decisions.

[0180] S808, Feedback adjustment: make adjustments based on the prediction results and verification status.

[0181] Adjustments are made based on the prediction results and verification conditions, and parameters are optimized to improve the consistency and reliability of battery performance.

[0182] In the present disclosure, the self-discharge prediction model is used to predict the battery self-discharge value, which can improve the prediction efficiency, accuracy and flexibility. It also has scalability and automation capabilities, can reduce costs, achieve real-time monitoring and anomaly detection, improve prediction performance through continuous learning and optimization, and provide data-driven decision support to help companies optimize production strategies and improve production efficiency and product quality.

[0183] Through machine learning technology, data processing in the battery production process has become more efficient and accurate. The model can quickly process large amounts of data, reducing manual operations and time consumption, thereby improving prediction efficiency. At the same time, the model can learn the complex patterns of the data to provide more accurate prediction results and improve prediction accuracy. In addition, machine learning models can easily handle nonlinear relationships and adapt to a variety of data types, enhancing the flexibility of the model.

[0184] The scalability and automation capabilities of the model also bring significant advantages. The model can be expanded to new data sets to adapt to changing business needs, improving the adaptability of the model. The model's automated prediction process reduces human intervention, improves the stability and consistency of predictions, and reduces costs. The model can monitor battery performance in real time, quickly respond to changes in the production process, and improve production efficiency. The model can detect outliers, which helps to detect problems in the production process at an early stage and improve production quality.

[0185] Based on the same inventive concept, the present disclosure also provides a battery self-discharge prediction device, as described in the following embodiments. Since the principle of solving the problem in the device embodiment is similar to that in the above method embodiment, the implementation of the device embodiment can refer to the implementation of the above method embodiment, and the repeated parts will not be repeated.

[0186] Fig.13 A schematic diagram of a battery self-discharge prediction device according to an embodiment of the present disclosure is shown. Fig.13 As shown, the device includes a data acquisition module 1310, a feature extraction module 1320, a data screening module 1330 and a model prediction module 1340.

[0187] The data acquisition module 1310 is used to acquire various process data of the battery to be tested during the production process, and the process data includes process parameter data of at least one of the production stage, the assembly stage and the formation stage;

[0188] A feature extraction module 1320 is used to extract features from various process data to obtain multiple process features and calculate the feature importance value of each process feature;

[0189] A data screening module 1330 is used to screen target process features whose feature importance values ​​meet preset conditions, and screen target process data corresponding to the target process features;

[0190] The model prediction module 1340 is used to process the target process data through the self-discharge prediction model to obtain the self-discharge prediction value of the battery to be tested, wherein the self-discharge prediction model is trained by various historical process data of battery samples and corresponding battery sample self-discharge labels.

[0191] In one embodiment, the device also includes a model training module not shown in the accompanying drawings, and the model training module is used to train the self-discharge prediction model in the following manner: determine multiple initial hyperparameter combinations of the self-discharge prediction model, the initial hyperparameter combinations are multiple data groups composed of different values ​​of multiple hyperparameters of the self-discharge prediction model; construct a data set based on multiple historical process data and corresponding battery sample self-discharge labels, the data set includes a training data set and a verification data set; traverse multiple initial hyperparameter combinations, and iteratively train the self-discharge prediction model to be trained based on the training data set until convergence, so as to obtain the self-discharge prediction model corresponding to each initial hyperparameter combination; process the verification data set through the self-discharge prediction model corresponding to each initial hyperparameter combination, and obtain the evaluation index score of the self-discharge prediction model corresponding to each initial hyperparameter combination; take the initial hyperparameter combination with the highest evaluation index score as the optimal hyperparameter combination, and take the self-discharge prediction model corresponding to the optimal hyperparameter combination as the trained self-discharge prediction model.

[0192] It should be noted that the hyperparameters of the self-discharge prediction model include at least one of the learning rate, the maximum depth of the decision tree, the number of decision trees, the minimum sample weight and of the leaf nodes, the sample sampling ratio of each decision tree, the feature sampling ratio of each decision tree, the minimum split loss, and the weight of the regularization term.

[0193] It should be noted that the evaluation index of the self-discharge prediction model includes at least one of mean square error, determination coefficient, accuracy, and recall rate.

[0194] In one embodiment, the model training module is further used to regularly update the data set based on multiple process data of the battery to be tested and the corresponding battery self-discharge labels, and use the updated data set to retrain the self-discharge prediction model.

[0195] In one embodiment, the data screening module 1330 is used to screen target process features whose feature importance values ​​meet preset conditions in any of the following ways: if the feature importance value is greater than or equal to a preset feature importance value threshold, then the feature importance value is determined to meet the preset conditions; sorting multiple process features in descending order based on the feature importance value, if the sorting order of the feature importance value is less than or equal to a first preset order threshold, then the feature importance value is determined to meet the preset conditions; sorting multiple process features in ascending order based on the feature importance value, if the sorting order of the feature importance value is greater than or equal to the preset order threshold, then the feature importance value is determined to meet the preset conditions.

[0196] In one embodiment, the self-discharge prediction model is an extreme gradient boosting model; wherein the feature extraction module is used to calculate the feature importance value of each process feature by any of the following methods: counting the number of times each process feature is used at a split node in all decision trees, and taking the number of times used as the feature importance value; calculating the information gain of each process feature, and taking the information gain as the feature importance value, the information gain represents the average reduction in the loss function value of each process feature when it is used as a split node; calculating the coverage of each process feature, and taking the coverage as the feature importance value, the coverage represents the average number of samples covered when each process feature is used as a split node in all decision trees.

[0197] It should be noted that the process characteristics include raw material ratio, stirring time, coating thickness, rolling density, cutting accuracy, battery cell size, electrode coating uniformity, pole piece alignment, winding tension, welding quality, sealing process, formation conditions, capacity separation test results, hot melt pressure, hot melt time, total baking time, preheating time, cathode water capacity, dielectric withstand voltage test time, electrolyte water content, electrolyte hydrofluoric acid, preset temperature standing time, formation starting voltage, formation capacity, end temperature, full welding power, capacity after load adjustment, capacity starting voltage, discharge capacity, charging capacity, standing time, open circuit voltage, open circuit voltage test temperature, battery cell temperature, helium inspection production time, vacuum baking production time, preset temperature standing production time, formation production time, capacity production time, room temperature standing production time, and open circuit voltage or insulation pulse test generation time, one or more of the time.

[0198] In one embodiment, the device also includes a data display module not shown in the accompanying drawings. The data display module is used to visually display the self-discharge prediction value and self-discharge label of the battery to be tested after processing the target process data through the self-discharge prediction model to obtain the self-discharge prediction value of the battery to be tested, so that the user can make production decisions based on the self-discharge prediction value and self-discharge label of the battery to be tested.

[0199] In the embodiment of the present disclosure, a variety of process data of the battery to be tested in the production process are obtained, and the process data include process parameter data of at least one of the film production stage, the assembly stage and the formation stage; feature extraction is performed on the various process data to obtain multiple process features, and the feature importance value of each process feature is calculated; the target process feature whose feature importance value meets the preset conditions is screened, and the target process data corresponding to the target process feature is screened; the target process data is processed by a self-discharge prediction model to obtain the self-discharge prediction value of the battery to be tested, and the self-discharge prediction model is obtained by training the various historical process data of the battery sample and the corresponding battery sample self-discharge label. On the one hand, the present disclosure processes the process data in the battery production process by the self-discharge prediction model, effectively improving the accuracy and prediction efficiency of the self-discharge prediction. On the other hand, by calculating the feature importance value of the process feature and screening the target process feature according to the feature importance value, effective process parameter information can be extracted, thereby screening the process feature that has a greater impact on the battery self-discharge, and further improving the accuracy of the battery self-discharge prediction.

[0200] Those skilled in the art will appreciate that various aspects of the present disclosure may be implemented as systems, methods or program products. Therefore, various aspects of the present disclosure may be specifically implemented in the following forms, namely: complete hardware implementation, complete software implementation (including firmware, microcode, etc.), or a combination of hardware and software, which may be collectively referred to herein as "circuits", "modules" or "systems".

[0201] Refer to the following Fig.14 1400 according to this embodiment of the present disclosure is described. Fig.14 The electronic device 1400 shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0202] In one embodiment, the electronic device 1400 includes a processor and a memory, where the memory is used to store executable instructions of the processor, wherein the processor is configured to perform the above-mentioned battery self-discharge prediction method by executing the executable instructions.

[0203] like Fig.14 As shown, the electronic device 1400 is in the form of a general computing device. The components of the electronic device 1400 may include but are not limited to: at least one processing unit 1410, at least one storage unit 1420, and a bus 1430 connecting different system components (including the storage unit 1420 and the processing unit 1410).

[0204] Among them, the storage unit stores a program code, and the program code can be executed by the processing unit 1410, so that the processing unit 1410 executes the steps described in the above "Exemplary Method" section of this specification according to various exemplary embodiments of the present disclosure. For example, the processing unit 1410 can execute the following steps of the above method embodiment: obtaining a variety of process data of the battery to be tested during the production process, the process data including process parameter data of at least one of the film production stage, the assembly stage and the formation stage; extracting features from the various process data to obtain a plurality of process features, and calculating the feature importance value of each process feature; screening the target process features whose feature importance values ​​meet the preset conditions, and screening the target process data corresponding to the target process features; processing the target process data through the self-discharge prediction model to obtain the self-discharge prediction value of the battery to be tested, and the self-discharge prediction model is obtained by training the various historical process data of the battery sample and the corresponding battery sample self-discharge label.

[0205] The storage unit 1420 may include a readable medium in the form of a volatile storage unit, such as a random access storage unit (RAM) 14201 and / or a cache storage unit 14202 , and may further include a read-only storage unit (ROM) 14203 .

[0206] The storage unit 1420 may also include a program / utility 14204 having a set (at least one) of program modules 14205, such program modules 14205 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0207] Bus 1430 may represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0208] The electronic device 1400 may also communicate with one or more external devices 1440 (e.g., keyboards, pointing devices, Bluetooth devices, etc.), one or more devices that enable a user to interact with the electronic device 1400, and / or any device that enables the electronic device 1400 to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed through an input / output (I / O) interface 1450. Furthermore, the electronic device 1400 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) through a network adapter 1460. Fig.14As shown, the network adapter 1460 communicates with other modules of the electronic device 1400 via the bus 1430. It should be understood that, although not shown in the figure, other hardware and / or software modules can be used in conjunction with the electronic device 1400, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0209] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the implementation of the present disclosure.

[0210] In particular, according to an embodiment of the present disclosure, the process described in the above reference flowchart can be implemented as a computer program product, which includes: a computer program, which, when executed by a processor, implements the steps described in the above "Exemplary Method" section according to various exemplary embodiments of the present disclosure.

[0211] In an exemplary embodiment of the present disclosure, a computer-readable storage medium is also provided, on which a computer program is stored, and when the computer program is executed by a processor, the steps according to various exemplary embodiments of the present disclosure described in the above "Exemplary Method" section are implemented. The computer-readable storage medium can be a readable signal medium or a readable storage medium.

[0212] In some possible implementations, various aspects of the present disclosure may also be implemented in the form of a program product, which includes a computer program, and when the computer program is executed by a processor, the battery self-discharge prediction method in any of the above embodiments is implemented. In a feasible embodiment, the program product includes a program code, and when the program product is run on a terminal device, the program code is used to enable the terminal device to execute the steps according to various exemplary embodiments of the present disclosure described in the above “Exemplary Method” section of this specification.

[0213] More specific examples of computer-readable storage media in the present disclosure may include, but are not limited to, an electrical connection having one or more conductors, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0214] In the present disclosure, a computer readable storage medium may include a data signal propagated in baseband or as part of a carrier wave, wherein a readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A readable signal medium may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0215] Alternatively, the program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the foregoing.

[0216] In a specific implementation, the program code for performing the operations of the present disclosure may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., using an Internet service provider to connect through the Internet).

[0217] It should be noted that, although several modules or units of the device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into multiple modules or units to be embodied.

[0218] In addition, although the steps of the method in the present disclosure are described in a specific order in the drawings, this does not require or imply that the steps must be performed in this specific order, or that all the steps shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps, etc.

[0219] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a mobile terminal, or a network device, etc.) to execute the method according to the implementation of the present disclosure.

[0220] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. The present disclosure is intended to cover any variations, uses or adaptations of the present disclosure, which follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The description and examples are intended to be exemplary only, and the true scope and spirit of the present disclosure are indicated by the appended claims.

Claims

1. A method for predicting battery self-discharge, characterized in that: include: Acquire a variety of process data of the battery to be tested during the production process, wherein the process data includes process parameter data of at least one of the film production stage, the assembly stage and the formation stage; Performing feature extraction on the multiple process data to obtain multiple process features, and calculating feature importance values ​​of each of the process features; Screening target process features whose feature importance values ​​meet preset conditions, and screening target process data corresponding to the target process features; The target process data is processed by a self-discharge prediction model to obtain a self-discharge prediction value of the battery to be tested, wherein the self-discharge prediction model is trained by various historical process data of battery samples and corresponding battery sample self-discharge labels.

2. The battery self-discharge prediction method according to claim 1, characterized in that: The self-discharge prediction model is trained in the following way: Determine a plurality of initial hyperparameter combinations of the self-discharge prediction model, wherein the initial hyperparameter combinations are a plurality of data groups formed by different values ​​of the plurality of hyperparameters of the self-discharge prediction model; Constructing a data set based on a variety of historical process data and corresponding battery sample self-discharge labels, wherein the data set includes a training data set and a verification data set; Traversing the multiple initial hyperparameter combinations, iteratively training the self-discharge prediction model to be trained based on the training data set until convergence, to obtain the self-discharge prediction model corresponding to each of the initial hyperparameter combinations; The validation data set is processed by the self-discharge prediction model corresponding to each of the initial hyperparameter combinations to obtain an evaluation index score of the self-discharge prediction model corresponding to each of the initial hyperparameter combinations; The initial hyperparameter combination with the highest evaluation index score is used as the optimal hyperparameter combination, and the self-discharge prediction model corresponding to the optimal hyperparameter combination is used as the trained self-discharge prediction model.

3. The battery self-discharge prediction method according to claim 2, characterized in that: The hyperparameters of the self-discharge prediction model include at least one of a learning rate, a maximum depth of a decision tree, the number of decision trees, a minimum sample weight and a leaf node, a sample sampling ratio of each decision tree, a feature sampling ratio of each decision tree, a minimum splitting loss, and a weight of a regularization term.

4. The battery self-discharge prediction method according to claim 2, characterized in that: The evaluation index of the self-discharge prediction model includes at least one of mean square error, determination coefficient, accuracy and recall rate.

5. The battery self-discharge prediction method according to claim 2, characterized in that: The method further comprises: Based on the multiple process data of the battery to be tested and the corresponding battery self-discharge labels, the data set is updated regularly, and the self-discharge prediction model is retrained using the updated data set.

6. The battery self-discharge prediction method according to claim 1, characterized in that: The screening of the target process features whose feature importance values ​​meet the preset conditions includes any of the following methods: If the feature importance value is greater than or equal to a preset feature importance value threshold, it is determined that the feature importance value meets the preset condition; Sorting the plurality of process features in descending order based on the feature importance values, and if the order of the feature importance values ​​is less than or equal to a first preset order threshold, determining that the feature importance values ​​meet a preset condition; Based on the feature importance values, a plurality of process features are sorted in ascending order. If the sorting order of the feature importance values ​​is greater than or equal to the preset order threshold, it is determined that the feature importance values ​​meet a preset condition.

7. The battery self-discharge prediction method according to claim 1, characterized in that: The self-discharge prediction model is an extreme gradient boosting model; The calculation of the feature importance value of each of the process features includes any one of the following: Counting the number of times each of the process features is used at the splitting nodes in all decision trees, and using the number of times used as the feature importance value; Calculating the information gain of each of the process features, and using the information gain as the feature importance value, wherein the information gain represents an average reduction in the loss function value of each of the process features when serving as a splitting node; The coverage of each of the process features is calculated and the coverage is used as the feature importance value. The coverage represents the average value of the number of samples covered when each of the process features is used as a split node in all decision trees.

8. The battery self-discharge prediction method according to claim 1, characterized in that: The process characteristics include raw material ratio, stirring time, coating thickness, rolling density, cutting accuracy, battery cell size, electrode coating uniformity, pole piece alignment, winding tension, welding quality, sealing process, formation conditions, capacity distribution test results, hot melt pressure, hot melt time, total baking time, preheating time, cathode water capacity, dielectric withstand voltage test time, electrolyte water content, electrolyte hydrofluoric acid, preset temperature standing time, formation starting voltage, formation capacity, end temperature, full welding power, capacity after load adjustment, capacity starting voltage, discharge capacity, charging capacity, standing time, open circuit voltage, open circuit voltage test temperature, battery cell temperature, helium inspection production time, vacuum baking production time, preset temperature standing production time, formation production time, capacity production time, room temperature standing production time, and open circuit voltage or insulation pulse test generation time. One or more of the time.

9. The battery self-discharge prediction method according to claim 1, characterized in that: After the target process data is processed by the self-discharge prediction model to obtain the self-discharge prediction value of the battery to be tested, the method further includes: The self-discharge prediction value and the self-discharge label of the battery to be tested are displayed visually, so that the user can make production decisions according to the self-discharge prediction value and the self-discharge label of the battery to be tested.

10. A battery self-discharge prediction device, characterized in that: include: A data acquisition module, used to acquire various process data of the battery to be tested during the production process, wherein the process data includes process parameter data of at least one of the film production stage, the assembly stage and the formation stage; A feature extraction module, used to extract features from the multiple process data to obtain multiple process features, and calculate the feature importance value of each of the process features; A data screening module, used to screen target process features whose feature importance values ​​meet preset conditions, and screen target process data corresponding to the target process features; The model prediction module is used to process the target process data through a self-discharge prediction model to obtain a self-discharge prediction value of the battery to be tested, wherein the self-discharge prediction model is obtained by training multiple historical process data of battery samples and corresponding battery sample self-discharge labels.

11. An electronic device, characterized in that: include: processor; as well as A memory, configured to store executable instructions of the processor; The processor is configured to execute the battery self-discharge prediction method according to any one of claims 1 to 9 by executing the executable instructions.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the battery self-discharge prediction method according to any one of claims 1 to 9 is implemented.

13. A computer program product, characterized in that The invention comprises a computer program, which, when executed by a processor, implements the battery self-discharge prediction method according to any one of claims 1 to 9.

Images