Battery parameter prediction method and apparatus, and device, storage medium and program product

Abstract

The present application relates to a battery parameter prediction method and apparatus, and a device, a storage medium and a program product. The method comprises: acquiring battery parameters of a battery; inputting the battery parameters into a rate prediction model, and determining a first predicted charging rate of the battery on the basis of an output of the rate prediction model; when a preset termination condition is not met, on the basis of a preset battery parameter constraint condition, determining at least one first candidate battery parameter and a second predicted charging rate corresponding to each first candidate battery parameter, and determining a target candidate battery parameter from the first candidate battery parameters on the basis of the first predicted charging rate and the second predicted charging rate corresponding to each first candidate battery parameter; and when the preset termination condition is met, determining an ideal design parameter for the battery on the basis of the target candidate battery parameter. The embodiments of the present application can improve the efficiency of research and development of batteries.

Description

Battery parameter prediction methods, devices, equipment, storage media, and software products

[0001] Cross-references

[0002] This application incorporates Chinese Patent Application No. 2024105652448, filed on May 8, 2024, entitled “Battery Parameter Prediction Method, Apparatus, Device, Storage Medium and Program Product”, which is incorporated herein by reference in its entirety. Technical Field

[0003] This application relates to the field of battery technology, and in particular to a method, apparatus, device, storage medium, and program product for predicting battery parameters. Background Technology

[0004] With the development of new energy technologies, the types and numbers of electrical devices that incorporate rechargeable batteries are increasing. Therefore, accelerating battery research and development is crucial for battery companies.

[0005] In related technologies, researchers typically predict and study battery parameters using electrochemical mechanism formulas based on information such as battery charging power and battery capacity. However, these technologies suffer from low battery development efficiency.

[0006] Summary of the Invention

[0007] In view of the above problems, this application provides a battery parameter prediction method, apparatus, device, storage medium and program product, which can solve the problem of low battery research and development efficiency in related technologies.

[0008] In a first aspect, this application provides a method for predicting battery parameters, the method comprising:

[0009] Obtain the battery parameters, including design parameters, test parameters, and process parameters;

[0010] The battery parameters are input into the rate prediction model, and the first predicted charging rate of the battery is determined based on the output of the rate prediction model.

[0011] If the preset termination condition is not met, at least one first candidate battery parameter and a second predicted charging rate corresponding to each first candidate battery parameter are determined based on preset battery parameter constraints; wherein, the preset battery parameter constraints include preset design parameter constraints, preset test parameter constraints and preset process parameter constraints.

[0012] Based on the first predicted charging rate and the second predicted charging rate corresponding to each first candidate battery parameter, the target candidate battery parameters are determined from each first candidate battery parameter.

[0013] Under the condition that the preset termination conditions are met, the ideal design parameters of the battery are determined based on the target candidate battery parameters.

[0014] Compared to related technologies that predict battery charging power and capacity using electrochemical mechanism formulas, this embodiment of the application uses a rate prediction model to predict the charging rate. This model considers battery design parameters, testing parameters, and process parameters, making it more comprehensive. Therefore, the method used in this embodiment not only has higher prediction efficiency but also more accurate predicted charging rates. Furthermore, by determining the target candidate battery parameters from the first candidate battery parameters based on the first predicted charging rate and the second predicted charging rate corresponding to each first candidate battery parameter, researchers can quickly obtain the ideal design parameters for achieving a specific charging rate. This significantly reduces the number and duration of battery experiments, thereby saving battery development costs and improving development efficiency.

[0015] In some embodiments, determining target candidate battery parameters from each first candidate battery parameter based on a first predicted charging rate and a second predicted charging rate corresponding to each first candidate battery parameter includes:

[0016] Based on the first predicted charging rate and the second predicted charging rate corresponding to each first candidate battery parameter, the second predicted charging rate when the function value of the first objective function is minimized is determined based on the first objective function.

[0017] The parameters of the first candidate battery corresponding to the second predicted charging rate are used as the target candidate battery parameters.

[0018] In this embodiment of the application, the target candidate battery parameters are determined from the first candidate battery parameters based on the first predicted charging rate and the second predicted charging rate corresponding to each first candidate battery parameter, according to the first objective function. This achieves the goal of parameter optimization for the target candidate battery parameters, so that more accurate ideal design parameters for the battery can be determined.

[0019] In some embodiments, the first objective function is a function that minimizes the absolute value of the difference between the first predicted charging rate and the charging rate predicted by the rate prediction model.

[0020] In some embodiments, determining at least one first candidate battery parameter and a second predicted charging rate corresponding to each first candidate battery parameter based on preset battery parameter constraints includes:

[0021] Based on preset battery parameter constraints, at least one first candidate battery parameter is generated according to a preset optimization algorithm.

[0022] The parameters of each first candidate battery are input into the rate prediction model to obtain the second predicted charging rate corresponding to each first candidate battery parameter.

[0023] In this embodiment, by combining preset battery parameter constraints and preset optimization algorithms to determine the first candidate battery parameters, more suitable first candidate battery parameters can be generated. This allows for the determination of more suitable target candidate battery parameters based on the more suitable first candidate battery parameters and the corresponding second predicted charging rate, thereby facilitating the acquisition of more accurate ideal design parameters.

[0024] In some embodiments, based on preset battery parameter constraints, at least one first candidate battery parameter is generated according to a preset optimization algorithm, including:

[0025] Based on preset battery parameter constraints, at least one first candidate battery parameter is randomly generated.

[0026] In some embodiments, based on preset battery parameter constraints, at least one first candidate battery parameter is generated according to a preset optimization algorithm, including:

[0027] Based on preset battery parameter constraints, at least one first candidate battery parameter is obtained through at least one iterative calculation according to the first preset Gaussian mixture model and the second preset Gaussian mixture model.

[0028] In some embodiments, based on preset battery parameter constraints, at least one first candidate battery parameter is obtained through at least one iteration calculation according to a first preset Gaussian mixture model and a second preset Gaussian mixture model, including:

[0029] For each iteration of the calculation, based on the preset battery parameter constraints, when the ratio of the first preset Gaussian mixture model to the second preset Gaussian mixture model is maximized, the design hyperparameters corresponding to the first preset Gaussian mixture model, as well as the test hyperparameters and process hyperparameters corresponding to the second preset Gaussian mixture model, are used as the first candidate battery parameters.

[0030] In some embodiments, the method further includes:

[0031] If the preset termination condition is not met, at least one second candidate battery parameter and a third predicted charging rate corresponding to each second candidate battery parameter are determined based on the preset battery parameter constraint conditions.

[0032] Based on the third predicted charging rate corresponding to each second candidate battery parameter, the candidate charging rate prediction boundary is determined from each third predicted charging rate based on the second objective function.

[0033] If the preset termination conditions are met, the candidate charging rate prediction boundary will be used as the target charging rate prediction boundary.

[0034] In this embodiment of the application, by determining the candidate charging rate prediction boundary from each third predicted charging rate based on the third predicted charging rate corresponding to each second candidate battery parameter and the second objective function, the charging rate prediction boundary of the rate prediction model can be explored. This makes it easier for R&D personnel to know the rate prediction boundary of the rate prediction model, so as to select a rate prediction model that is more suitable for actual needs.

[0035] In some embodiments, if the second objective function is a function that maximizes the charging rate predicted by the rate prediction model, the candidate charging rate prediction boundary is determined from each of the third predicted charging rates based on the second objective function, according to the third predicted charging rate corresponding to each second candidate battery parameter, including:

[0036] The maximum value among the third predicted charging rates corresponding to each of the second candidate battery parameters is taken as the upper boundary of the candidate charging rate prediction.

[0037] In some embodiments, if the second objective function is a function that minimizes the charging rate predicted by the rate prediction model, the candidate charging rate prediction boundary is determined from each of the third predicted charging rates based on the second objective function, according to the third predicted charging rate corresponding to each second candidate battery parameter, including:

[0038] The minimum value among the third predicted charging rates corresponding to each of the second candidate battery parameters is taken as the lower boundary of the candidate charging rate prediction.

[0039] In some embodiments, determining at least one second candidate battery parameter and a third predicted charging rate corresponding to each second candidate battery parameter based on preset battery parameter constraints includes:

[0040] Based on preset battery parameter constraints, at least one second candidate battery parameter is generated according to a preset optimization algorithm.

[0041] The parameters of each second candidate battery are input into the rate prediction model to obtain the third predicted charging rate corresponding to each second candidate battery parameter.

[0042] In some embodiments, the method further includes:

[0043] Obtain the training sample set; wherein, each training sample in the training sample set includes: training battery parameters and corresponding training charging rates; the training battery parameters include training design parameters, training test parameters and training process parameters;

[0044] The initial prediction model is trained based on the training sample set to obtain the multiplier prediction model.

[0045] In this embodiment, the training process of the initial prediction model considers information such as battery training design parameters, training test parameters, and training process parameters. The factors considered are more comprehensive. Therefore, the rate prediction model trained in this embodiment has higher accuracy, which allows for rapid and accurate prediction of the battery's predicted charging rate, thereby improving the efficiency of charging rate prediction.

[0046] In some embodiments, obtaining the training sample set includes:

[0047] Multiple sets of initial samples were obtained, including initial battery parameters and corresponding initial charging rates; the initial battery parameters included initial design parameters, initial test parameters, and initial process parameters.

[0048] Data preprocessing is performed on multiple initial samples to obtain a training sample set.

[0049] In this embodiment of the application, by preprocessing the acquired multiple sets of initial samples, the data quality of the training samples can be improved, which is beneficial to improving the accuracy of the ratio prediction model, and thus improving the prediction precision of the ratio prediction model.

[0050] In some embodiments, obtaining multiple sets of initial samples includes:

[0051] Obtain multiple initial design parameters and their corresponding design identification information;

[0052] Based on multiple design identifiers, obtain the initial test parameters, initial process parameters, and initial charging rate corresponding to each design identifier;

[0053] The initial design parameters, initial test parameters, initial process parameters, and initial charging rates corresponding to multiple design identification information are merged into multiple initial samples.

[0054] In some embodiments, data preprocessing is performed on multiple sets of initial samples to obtain a training sample set, including:

[0055] Based on the preset battery mechanism information, the initial battery parameters and corresponding initial charging rates in multiple initial samples are screened to obtain the first intermediate battery parameters and corresponding first intermediate charging rates in multiple first intermediate samples whose influence on the charging rate is greater than the preset influence level.

[0056] The correlation between the first intermediate battery parameters and the charging rate in multiple sets of first intermediate samples was determined.

[0057] Based on the correlation of the first intermediate battery parameters in multiple sets of first intermediate samples, the first intermediate battery parameters in multiple sets of first intermediate samples are recursively deleted to obtain multiple sets of training samples with a correlation greater than the preset correlation with the charging rate.

[0058] In this application, by combining the basic knowledge of electrochemical mechanisms with statistical theory, several design parameters, test parameters, and process parameters that most affect the charging rate are selected. This not only improves the accuracy of the rate prediction model but also reduces the computational cost in the modeling and use process.

[0059] In some embodiments, data preprocessing is performed on multiple sets of initial samples to obtain a training sample set, including:

[0060] Based on the synchronization fault information of the database used to store battery information, the initial battery parameters and corresponding initial charging rates in multiple initial samples are subjected to the first data filling process to obtain the second intermediate battery parameters and corresponding second intermediate charging rates in multiple second intermediate samples.

[0061] According to the preset filling method, the second intermediate battery parameters and the corresponding second intermediate charging rate in multiple sets of second intermediate samples are subjected to second data filling processing to obtain multiple sets of training samples.

[0062] In this embodiment of the application, by combining the filling method of synchronous fault information in the database with the preset filling method, the data integrity of the training samples can be improved, thereby improving the data quality of the training samples and helping to improve the accuracy of the ratio prediction model.

[0063] In some embodiments, data preprocessing is performed on multiple sets of initial samples to obtain a training sample set, including:

[0064] Based on the preset battery mechanism information, the data that exceed the corresponding preset parameter boundary in the initial battery parameters and corresponding initial charging rates of multiple initial samples are removed to obtain the third intermediate battery parameters and third intermediate charging rates in multiple third intermediate samples.

[0065] Based on different test conditions, the third intermediate battery parameters and corresponding third intermediate charging rates in multiple sets of third intermediate samples are divided into multiple sets of intermediate parameters; each set of intermediate parameters includes multiple third intermediate battery parameters and corresponding third intermediate charging rates.

[0066] Calculate the mean absolute percentage error within each of the multiple intermediate parameter sets, and remove data whose mean absolute percentage error within each set is greater than the preset error to obtain multiple training samples.

[0067] In this application, by combining preset parameter boundaries and mean absolute percentage error for data cleaning, the accuracy of training sample data can be improved, thereby improving the data quality of training samples and thus improving the accuracy of the ratio prediction model.

[0068] Secondly, this application provides a battery parameter prediction device, the device comprising:

[0069] The module is used to acquire battery parameters, which include design parameters, test parameters, and process parameters.

[0070] The first determining module is used to input battery parameters into the rate prediction model and determine the first predicted charging rate of the battery based on the output of the rate prediction model.

[0071] The second determining module is used to determine at least one first candidate battery parameter and a second predicted charging rate corresponding to each first candidate battery parameter based on preset battery parameter constraints when the preset termination condition is not met; wherein the preset battery parameter constraints include preset design parameter constraints, preset test parameter constraints and preset process parameter constraints.

[0072] The third determining module is used to determine the target candidate battery parameters from each first candidate battery parameter based on the first predicted charging rate and the second predicted charging rate corresponding to each first candidate battery parameter.

[0073] The fourth determination module is used to determine the ideal design parameters of the battery based on the target candidate battery parameters, provided that the preset termination conditions are met.

[0074] Thirdly, this application provides a battery parameter prediction device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of any of the methods in the first aspect above.

[0075] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the method described in any of the first aspects above.

[0076] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the method described in any of the first aspects above.

[0077] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0078] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiments below. The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of this application. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0079] Figure 1 is a schematic diagram of the application environment provided in the embodiments of this application;

[0080] Figure 2 is a flowchart illustrating a battery parameter prediction method provided in some embodiments of this application;

[0081] Figure 3 is a flowchart illustrating a battery parameter prediction method provided in some other embodiments of this application;

[0082] Figure 4 is a flowchart illustrating a battery parameter prediction method provided in some other embodiments of this application;

[0083] Figure 5 is a flowchart illustrating a battery parameter prediction method provided in some other embodiments of this application;

[0084] Figure 6 is a schematic diagram showing the importance of each feature parameter corresponding to the ratio prediction model provided in some embodiments of this application;

[0085] Figure 7 is a thermodynamic diagram showing the correlation between various characteristic parameters of the rate prediction model provided in some embodiments of this application.

[0086] Figure 8 is a flowchart illustrating the training method of the ratio prediction model provided in some embodiments of this application;

[0087] Figure 9 is a flowchart illustrating the method for obtaining training sample sets provided in some embodiments of this application;

[0088] Figure 10 is a schematic diagram of the battery parameter prediction model provided in the embodiment of this application;

[0089] Figure 11 is a schematic diagram of the prediction accuracy of the ratio prediction model provided in the embodiment of this application;

[0090] Figure 12 is a schematic diagram of the prediction accuracy of the ratio prediction model provided in the embodiments of this application (II).

[0091] Figure 13 is a schematic diagram of the iteration of the objective function value of the random search algorithm provided in the embodiment of this application;

[0092] Figure 14 is a schematic diagram of the difference frequency histogram of the objective function values ​​of the random search algorithm provided in the embodiment of this application;

[0093] Figure 15 is a schematic diagram of the iteration of the objective function value of the simulated annealing algorithm provided in the embodiment of this application;

[0094] Figure 16 is a schematic diagram of the difference frequency histogram of the objective function value of the simulated annealing algorithm provided in the embodiment of this application;

[0095] Figure 17 is a schematic diagram of the iteration of the objective function value of the TPE algorithm provided in the embodiment of this application;

[0096] Figure 18 is a schematic diagram of the difference frequency histogram of the objective function value of the TPE algorithm provided in the embodiment of this application;

[0097] Figure 19 is a schematic diagram of the iteration of the objective function value of the random search algorithm provided in the embodiment of this application;

[0098] Figure 20 is a schematic diagram of the difference frequency histogram of the objective function values ​​of the random search algorithm provided in the embodiment of this application;

[0099] Figure 21 is a schematic diagram of the iteration of the objective function value of the simulated annealing algorithm provided in the embodiment of this application;

[0100] Figure 22 is a schematic diagram of the difference frequency histogram of the objective function values ​​of the simulated annealing algorithm provided in the embodiment of this application;

[0101] Figure 23 is a schematic diagram of the iteration of the objective function value of the TPE algorithm provided in the embodiment of this application;

[0102] Figure 24 is a schematic diagram of the difference frequency histogram of the objective function value of the TPE algorithm provided in the embodiment of this application;

[0103] Figure 25 is a schematic diagram of the iteration of the objective function value of the random search algorithm provided in the embodiment of this application;

[0104] Figure 26 is a schematic diagram of the difference frequency histogram of the objective function values ​​of the random search algorithm provided in the embodiment of this application;

[0105] Figure 27 is a schematic diagram of the iteration of the objective function value of the simulated annealing algorithm provided in the embodiment of this application;

[0106] Figure 28 is a schematic diagram of the difference frequency histogram of the objective function values ​​of the simulated annealing algorithm provided in the embodiment of this application;

[0107] Figure 29 is a schematic diagram of the iteration of the objective function value of the TPE algorithm provided in the embodiment of this application;

[0108] Figure 30 is a schematic diagram of the difference frequency histogram of the objective function value of the TPE algorithm provided in the embodiment of this application;

[0109] Figure 31 is a schematic diagram of the structure of a battery parameter prediction device provided in some embodiments of this application;

[0110] Figure 32 is a schematic diagram of the structure of a battery parameter prediction device provided in some embodiments of this application. Detailed Implementation

[0111] The embodiments of the technical solution of this application will now be described in detail with reference to the accompanying drawings. These embodiments are only used to more clearly illustrate the technical solution of this application and are therefore merely examples, and should not be used to limit the scope of protection of this application.

[0112] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application; the term "comprising" and any variations thereof in the specification, claims and foregoing description of the drawings are intended to cover non-exclusive inclusion.

[0113] In the description of the embodiments of this application, technical terms such as "first" and "second" are used only to distinguish different objects and should not be construed as indicating or implying relative importance or implicitly specifying the number, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, "multiple" means two or more (including two), unless otherwise explicitly defined.

[0114] The battery parameter prediction method, apparatus, device, storage medium, and program products provided in this application can be applied to battery research and development application scenarios; of course, they can also be applied to other application scenarios.

[0115] With governments worldwide introducing tax incentives, industry benefits, and subsidies for new energy vehicle purchases, the new energy sector is experiencing rapid growth. However, this rapid development is largely due to the rapid increase in new energy vehicle sales; still, the total number of new energy vehicles remains relatively small compared to gasoline-powered vehicles.

[0116] With the continued rapid development of the economy, consumer demand for cars will remain strong. However, the emergence of electric vehicles in recent years has also brought some pain points and challenges to both users and automakers. For example, range anxiety, insufficient charging infrastructure, and long charging times deter some potential buyers. Furthermore, the long R&D and testing cycles and high R&D costs have led to a decline in the competitiveness of companies. Therefore, accelerating the R&D and iteration of electric vehicle batteries is not only crucial for enhancing the competitiveness of companies but also for alleviating user anxiety, making it a significant undertaking.

[0117] Batteries are the core component of electric vehicles, and predicting battery charging rate performance is crucial to alleviating users' range anxiety. Most methods for predicting battery charging rate in related technologies are based on electrochemical mechanism formulas. However, this method is overly idealistic and fails to consider the influence of other parameters on the battery charging rate, resulting in inaccurate predictions and low prediction efficiency. Furthermore, in battery development, researchers often need to conduct extensive experiments to summarize the battery parameters corresponding to a specific charging rate in order to improve battery parameters. However, the efficiency of summarizing battery parameters in related technologies is low. Therefore, related technologies suffer from low battery development efficiency.

[0118] To address the low R&D efficiency of batteries in related technologies, this application proposes a method of predicting charging rates by inputting battery parameters into a rate prediction model. Since the rate prediction model considers battery design parameters, testing parameters, and process parameters, it takes into account a more comprehensive range of factors. Therefore, the method used in this application not only has higher prediction efficiency but also more accurate predicted charging rates. Furthermore, by determining the target candidate battery parameters from the first candidate battery parameters based on the first predicted charging rate and the second predicted charging rate corresponding to each first candidate battery parameter, researchers can quickly obtain the ideal design parameters for achieving a specific charging rate. This significantly reduces the number and duration of battery experiments conducted by researchers, thereby saving battery R&D costs and further improving R&D efficiency.

[0119] Figure 1 is a schematic diagram of the application environment provided in this embodiment. As shown in Figure 1, the terminal 102 can communicate with the server 104 via a network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network servers. The terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, and tablets. The server 104 can be implemented using a standalone server or a server cluster composed of multiple servers.

[0120] For example, terminal 102 can perform prediction using the battery parameter prediction method provided in this application embodiment when it detects that preset detection conditions are met. The preset detection conditions may include, but are not limited to, every preset time interval, or receiving a parameter prediction instruction. It should be understood that in the scenario of terminal prediction, terminal 102 can obtain a trained rate prediction model from server 104; of course, if the processing power of terminal 102 is sufficient, terminal 102 can train its own rate prediction model.

[0121] In another example, when the terminal 102 receives a parameter prediction instruction, it can send information such as battery parameters carried in the parameter prediction instruction to the server 104, so that the server 104 can make a prediction using the battery parameter prediction method provided in this application embodiment and return the prediction result to the terminal 102.

[0122] In another example, server 104 may use the battery parameter prediction method provided in this application embodiment to make a prediction when it detects that the preset detection conditions are met, and return the prediction result to terminal 102.

[0123] In some embodiments, FIG2 is a schematic flowchart of a battery parameter prediction method provided in some embodiments of this application. This application describes the method applied to a battery parameter prediction device as an example. The battery parameter prediction device may include, but is not limited to, the terminal or server shown in FIG1. ​​As shown in FIG2, the method of this application embodiment may include the following steps:

[0124] Step S201: Obtain the battery parameters.

[0125] For example, the battery parameters in the embodiments of this application may include, but are not limited to, design parameters, test parameters, and process parameters.

[0126] The design parameters in the embodiments of this application may refer to the battery design information. For example, the design parameters may include, but are not limited to, at least one of the following: upper main material, lower main material, platform voltage, and electrolyte.

[0127] The test parameters in this application embodiment may refer to battery test condition information. For example, test parameters may include, but are not limited to, at least one of the following: test temperature, test rate, and test time.

[0128] The process parameters in this application embodiment may refer to battery manufacturing process information. For example, process parameters may include, but are not limited to, formation voltage and / or formation temperature.

[0129] In this step, the battery parameter prediction device can obtain the battery parameters, which may include, but are not limited to, design parameters, test parameters, and process parameters.

[0130] In one possible implementation, the battery parameter prediction device can obtain the battery parameters by receiving parameter prediction instructions, wherein the parameter prediction instructions may include, but are not limited to, the battery parameters.

[0131] In another possible implementation, the battery parameter prediction device can receive battery parameters sent by other devices.

[0132] In another possible implementation, the battery parameter prediction device can obtain the battery parameters from a database.

[0133] Of course, battery parameter prediction devices can also obtain battery parameters through other means.

[0134] Step S202: Input the battery parameters into the rate prediction model, and determine the first predicted charging rate of the battery based on the output of the rate prediction model.

[0135] The rate prediction model in this application embodiment can be used to indicate the relationship between the battery parameters of different batteries and the corresponding predicted charging rate.

[0136] For example, the ratio prediction model in the embodiments of this application can be a machine learning model, or it can be a mathematical model.

[0137] In this step, the battery parameter prediction device can input the battery parameters into the rate prediction model and determine the first predicted charging rate of the battery based on the output of the rate prediction model.

[0138] In one possible implementation, the battery parameter prediction device can use the output of the rate prediction model as the first predicted charging rate of the battery.

[0139] In another possible implementation, the battery parameter prediction device can obtain the first predicted charging rate of the battery by performing preset data processing on the output of the rate prediction model.

[0140] It should be noted that, compared with the method of predicting based on battery charging power and battery capacity through electrochemical mechanism formulas in related technologies, the rate prediction model in this application embodiment considers information such as battery design parameters, test parameters and process parameters. It considers more comprehensive factors. Therefore, the method of predicting by inputting battery parameters into the rate prediction model in this application embodiment not only has higher prediction efficiency, but also predicts a more accurate charging rate.

[0141] Step S203: If the preset termination condition is not met, determine at least one first candidate battery parameter and the second predicted charging rate corresponding to each first candidate battery parameter based on the preset battery parameter constraint condition.

[0142] For example, the preset termination conditions involved in the embodiments of this application may include, but are not limited to, any one of the following: the number of iterations reaches a preset number of iterations, or the iteration duration reaches a preset iteration duration.

[0143] For example, the first candidate battery parameters involved in the embodiments of this application may include, but are not limited to, at least one of the first candidate design parameters, first candidate test parameters, and first candidate process parameters.

[0144] The preset battery parameter constraints involved in this application embodiment can be used to indicate the value range of each candidate battery parameter, so as to facilitate optimization within the corresponding value range. For example, the preset battery parameter constraints may include, but are not limited to: preset design parameter constraints, preset test parameter constraints, and preset process parameter constraints. Specifically, the preset design parameter constraints can be used to indicate the value range of each candidate design parameter; the preset test parameter constraints can be used to indicate the value range of each candidate test parameter; and the preset process parameter constraints can be used to indicate the value range of each candidate process parameter.

[0145] In one possible implementation, the battery parameter prediction device can determine preset battery parameter constraints based on the maximum and minimum values ​​of the corresponding training and testing parameters in the training and testing dataset corresponding to the rate prediction model. For example, the battery parameter prediction device can use the maximum value of the training and testing design parameter in the training and testing dataset corresponding to the rate prediction model as the upper bound of the value range in the preset design parameter constraints, and use the minimum value of the training and testing design parameter in the training and testing dataset corresponding to the rate prediction model as the lower bound of the value range in the preset design parameter constraints.

[0146] For example, a battery parameter prediction device can use the maximum value of the training and testing process parameters in the training and testing dataset corresponding to the rate prediction model as the upper bound of the value range in the preset process parameter constraints, and use the minimum value of the training and testing process parameters in the training and testing dataset corresponding to the rate prediction model as the lower bound of the value range in the preset process parameter constraints.

[0147] In another possible implementation, the battery parameter prediction device can obtain preset battery parameter constraints from other devices.

[0148] Of course, battery parameter prediction devices can also obtain preset battery parameter constraints through other means.

[0149] In this step, if the preset termination condition is not met, the battery parameter prediction device can determine at least one first candidate battery parameter and the second predicted charging rate corresponding to each first candidate battery parameter based on the preset battery parameter constraints.

[0150] In one possible implementation, the battery parameter prediction device can determine at least one first candidate battery parameter based on preset battery parameter constraints, and then determine the second predicted charging rate corresponding to each first candidate battery parameter according to the at least one first candidate battery parameter and the rate prediction model.

[0151] In another possible implementation, the battery parameter prediction device can determine at least one first candidate battery parameter and a second predicted charging rate corresponding to each first candidate battery parameter by means of preset battery parameter constraints and preset battery parameter rate prediction model; wherein, the preset battery parameter rate prediction model can be used to generate at least one first candidate battery parameter and a second predicted charging rate corresponding to each first candidate battery parameter based on preset battery parameter constraints.

[0152] Of course, the battery parameter prediction device can also determine at least one first candidate battery parameter and the second predicted charging rate corresponding to each first candidate battery parameter in other ways based on preset battery parameter constraints.

[0153] Step S204: Determine the target candidate battery parameters from the first candidate battery parameters based on the first predicted charging rate and the second predicted charging rate corresponding to each first candidate battery parameter.

[0154] In this step, the battery parameter prediction device determines the target candidate battery parameters from the first candidate battery parameters based on the first predicted charging rate and the second predicted charging rate corresponding to each first candidate battery parameter.

[0155] In one possible implementation, the battery parameter prediction device can determine the target candidate battery parameter from the first candidate battery parameters based on a first objective function, according to the first predicted charging rate and the second predicted charging rate corresponding to each first candidate battery parameter.

[0156] The first objective function in this application embodiment can be used to indicate the relationship between the first predicted charging rate and the charging rate predicted by the rate prediction model.

[0157] For example, the first objective function can be a function that takes the minimum value of the absolute value of the difference between the first predicted charging rate and the charging rate predicted by the rate prediction model.

[0158] It should be understood that the decision variables of the first objective function may include, but are not limited to, the variables corresponding to the input parameters of the rate prediction model. For example, the decision variables of the first objective function may include battery design parameters, testing parameters, and process parameters.

[0159] For example, the first objective function can be expressed as the following formula (1): f1=min|CR P -P(x,y,z)| (1)

[0160] Where f1 represents the first objective function. CR PP(x,y,z) represents the first predicted charging rate. P(x,y,z) represents the charging rate predicted by the rate prediction model, where x represents the design parameter variable, y represents the test parameter variable, and z represents the process parameter variable. It should be understood that for any first candidate battery parameter, when the battery parameter prediction device substitutes the first candidate design parameter into the design parameter variable x, the first candidate test parameter into the test parameter variable y, and the first candidate process parameter into the process parameter variable z, P(x,y,z) can represent the second predicted charging rate corresponding to that first candidate battery parameter.

[0161] Of course, the first objective function can also be expressed as other variations or equivalent formulas of the above formula (1).

[0162] Another example is that the first objective function can be a function that takes the minimum value of the absolute value of the difference between the charging rate predicted by the rate prediction model and the first predicted charging rate.

[0163] Of course, the first objective function can also be other forms of relationship function used to indicate the relationship between the first predicted charging rate and the charging rate predicted by the rate prediction model.

[0164] For ease of understanding, in the following embodiments of this application, the relevant content of determining the target candidate battery parameters from each first candidate battery parameter based on the first objective function is illustrated by taking "the first objective function is a function that takes the minimum value of the absolute value of the difference between the first predicted charging rate and the charging rate predicted by the rate prediction model" as an example.

[0165] For example, the battery parameter prediction device can determine the second predicted charging rate when the function value of the first objective function is minimized based on the first predicted charging rate and the second predicted charging rate corresponding to each first candidate battery parameter, and use the first candidate battery parameter corresponding to the second predicted charging rate as the target candidate battery parameter.

[0166] In this embodiment, the battery parameter prediction device can calculate the absolute value of the difference between the first predicted charging rate and the second predicted charging rate corresponding to each first candidate battery parameter based on a first objective function, and determine the second predicted charging rate with the smallest absolute value. Furthermore, the battery parameter prediction device can use the first candidate battery parameter corresponding to the second predicted charging rate with the smallest absolute value as the target candidate battery parameter, achieving the parameter optimization purpose of the target candidate battery parameter, so as to determine more accurate ideal design parameters for the battery.

[0167] It should be understood that, under the condition of satisfying a preset termination condition, the battery parameter prediction device can, based on a first objective function, determine the second predicted charging rate with the smallest absolute value according to a first predicted charging rate and the second predicted charging rates corresponding to multiple first candidate battery parameters, and use the first candidate battery parameter corresponding to the second predicted charging rate with the smallest absolute value as the target candidate battery parameter. Alternatively, the battery parameter prediction device can, under the condition of not satisfying the preset termination condition, determine the second predicted charging rate with the smallest absolute value according to a first objective function based on a first objective function according to a first predicted charging rate and the second predicted charging rates corresponding to each first candidate battery parameter, and use the first candidate battery parameter corresponding to the second predicted charging rate with the smallest absolute value as the latest target candidate battery parameter, and so on, until the latest target candidate battery parameter can be used as the final target candidate battery parameter under the condition of satisfying the preset termination condition.

[0168] It should be noted that the solution acceptance criteria for the optimization process involved in this application embodiment may include, but are not limited to: an iterative solution is generated in each loop iteration process; if the iterative solution is better than the current solution, the iterative solution can be used as the current solution (or the latest solution); if the iterative solution is better than the best solution to date, the iterative solution can be used as the best solution.

[0169] In another possible implementation, the battery parameter prediction device can determine the target candidate battery parameters from the first candidate battery parameters by inputting the first predicted charging rate and the second predicted charging rate corresponding to each first candidate battery parameter into a preset screening model.

[0170] Of course, the battery parameter prediction device can also determine the target candidate battery parameters from the first candidate battery parameters in other ways based on the first predicted charging rate and the second predicted charging rate corresponding to each first candidate battery parameter.

[0171] Step S205: Under the condition that the preset termination condition is met, determine the ideal design parameters of the battery based on the target candidate battery parameters.

[0172] In this step, if the preset termination conditions are met, the battery parameter prediction device can determine the ideal design parameters of the battery based on the target candidate battery parameters. The ideal design parameters of the battery can refer to the optimal design parameters when the battery reaches the first predicted charging rate.

[0173] For example, a battery parameter prediction device can use a target candidate design parameter from the target candidate battery parameters as the ideal design parameter for the battery.

[0174] Of course, the battery parameter prediction device can also use the target candidate design parameters in the target candidate battery parameters as the ideal design parameters of the battery, and / or use the target candidate process parameters in the target candidate battery parameters as the ideal process parameters of the battery.

[0175] It should be understood that, typically, multiple design parameters of a battery may correspond to the same charging rate. In this embodiment, by determining the target candidate battery parameters from among the first candidate battery parameters based on a first predicted charging rate and a second predicted charging rate corresponding to each first candidate battery parameter, researchers can quickly obtain the ideal design parameters corresponding to a desired charging rate. This significantly reduces the number and duration of battery experiments conducted by researchers, saving battery development costs and improving development efficiency, thereby enhancing corporate competitiveness and reducing costs while increasing efficiency.

[0176] In this embodiment, battery parameters are obtained and input into a rate prediction model. A first predicted charging rate is determined based on the model's output. Battery parameters include design parameters, test parameters, and process parameters. Further, if a preset termination condition is not met, at least one first candidate battery parameter and a second predicted charging rate corresponding to each first candidate battery parameter are determined based on preset battery parameter constraints. A target candidate battery parameter is then determined from the first candidate battery parameters based on the first predicted charging rate and the second predicted charging rate corresponding to each first candidate battery parameter. The preset battery parameter constraints include preset design parameter constraints, preset test parameter constraints, and preset process parameter constraints. If the preset termination condition is met, the ideal design parameters of the battery are determined based on the target candidate battery parameter. As can be seen, compared to the method of predicting battery charging power and capacity based on electrochemical mechanism formulas in related technologies, the method of predicting by inputting battery parameters into a rate prediction model in this embodiment of the application is more comprehensive in its consideration of factors, as the rate prediction model takes into account battery design parameters, test parameters, and process parameters. Therefore, the method of prediction using a rate prediction model in this embodiment of the application is not only more efficient, but also more accurate in predicting the charging rate. In addition, by determining the target candidate battery parameters from the first candidate battery parameters based on the first predicted charging rate and the second predicted charging rate corresponding to each first candidate battery parameter, researchers can quickly obtain the ideal design parameters corresponding to a certain charging rate. This can greatly reduce the number and time of battery experiments conducted by researchers, thereby saving battery development costs and improving battery development efficiency.

[0177] In some embodiments, the battery parameter prediction device of this application can also output the ideal design parameters of the battery. It should be understood that, in cases where the battery parameter prediction device also obtains the ideal test parameters and / or ideal process parameters of the battery, the battery parameter prediction device can also output the ideal test parameters and / or ideal process parameters of the battery.

[0178] In this embodiment, the ideal battery parameters are output so that researchers can view them and use them as a reference for battery development. These ideal battery parameters may include, but are not limited to, at least one of the following: ideal design parameters, ideal test parameters, and ideal process parameters.

[0179] In some embodiments, Figure 3 is a flowchart illustrating a battery parameter prediction method provided in other embodiments of this application. This application provides an exemplary description of the relevant content in step S203 above, which involves "determining at least one first candidate battery parameter and the second predicted charging rate corresponding to each first candidate battery parameter based on preset battery parameter constraints." As shown in Figure 3, the method of this application embodiment may include the following steps:

[0180] Step S2031: Based on the preset battery parameter constraints, generate at least one first candidate battery parameter according to the preset optimization algorithm.

[0181] The preset optimization algorithms involved in the embodiments of this application may include, but are not limited to, heuristic algorithms or exact algorithms.

[0182] For ease of understanding, the following embodiments of this application provide a brief introduction and explanation of the exact algorithm and the heuristic algorithm.

[0183] Exact algorithms are a type of algorithm for finding optimal solutions to complex problems. They involve accurately describing and modeling the problem and then using mathematical methods to find the optimal solution. Compared to heuristic algorithms, exact algorithms guarantee an optimal solution, but they typically have higher time complexity. Exact algorithms are widely used in computer science, operations research, mathematics, and other fields, such as shortest path algorithms in graph theory and the simplex method in linear programming.

[0184] Heuristic algorithms are computational methods commonly used to solve complex optimization problems. Their main idea is to simulate the wisdom and experience inherent in humans or nature to find the optimal solution. Compared to traditional mathematical methods, heuristic algorithms focus more on searching within an approximate solution space, thus enabling them to quickly find better results. There are many types of heuristic algorithms, some common ones including genetic algorithms, fish swarm algorithms, ant colony algorithms, particle swarm algorithms, and so on. These algorithms provide different mechanisms to solve different problems and generally possess good adaptability and scalability.

[0185] Generally, among these solution algorithms, if the problem to be solved is small in scale and the parameter optimization model can be modeled as a linear programming model, then an exact algorithm can be used to solve it. If the problem is large in scale and the parameter optimization model has some nonlinear parts that cannot be handled, or if it is a black box problem where the relationship between decision variables is difficult to describe, then it is best to use a heuristic algorithm to solve it. This can ensure both the speed of the solution and the obtaining of an approximate optimal solution.

[0186] For example, the heuristic algorithms involved in the embodiments of this application may include, but are not limited to, any of the following: random search algorithm, simulated annealing algorithm, and TPE algorithm.

[0187] The random search algorithm involves randomly selecting sample points within the feasible region each time a solution is generated, and then substituting these selected sample points into the objective function to calculate its value. After multiple selections of sample points and calculations of the objective function, the greedy selection process yields the sample points that maximize the objective function value, thus providing the optimal solution and its objective function.

[0188] Simulated annealing algorithm: This algorithm applies the atomic annealing process to solving combinatorial optimization problems. Starting from a relatively high initial temperature, the simulated annealing algorithm randomly searches for the global optimum of the objective function in the solution space as the temperature gradually decreases.

[0189] The Tree-Based Optimization (TPE) algorithm is a tree-based Bayesian optimization algorithm primarily used to solve global optimization problems of black-box functions. In each trial, for each hyperparameter, TPE maintains a Gaussian mixture model l(x) for the hyperparameters associated with the optimal objective value and another Gaussian mixture model g(x) for the remaining hyperparameters. The hyperparameter that maximizes l(x) / g(x) is selected as the next set of search values. The main advantage of the TPE algorithm is its ability to adaptively adjust the size of the parameter search space and find the global optimum in as few iterations as possible.

[0190] In this step, the battery parameter prediction device can generate at least one first candidate battery parameter based on preset battery parameter constraints and according to a preset optimization algorithm.

[0191] For example, the battery parameter prediction device can synchronously generate multiple first candidate battery parameters based on preset battery parameter constraints and according to a preset optimization algorithm. In another example, the battery parameter prediction device can cyclically generate first candidate battery parameters based on preset battery parameter constraints and according to a preset optimization algorithm, wherein each generated first candidate battery parameter can be stored separately, or the latest first candidate battery parameter can replace a historical first candidate battery parameter.

[0192] In one possible implementation, at least one first candidate battery parameter is randomly generated based on preset battery parameter constraints.

[0193] In this implementation, when the preset optimization algorithm is a random search algorithm or a simulated annealing algorithm, the battery parameter prediction device can randomly generate at least one first candidate battery parameter based on preset battery parameter constraints.

[0194] In another possible implementation, based on preset battery parameter constraints, at least one first candidate battery parameter is obtained through at least one iterative calculation according to a first preset Gaussian mixture model and a second preset Gaussian mixture model.

[0195] In this implementation, when the preset optimization algorithm is the TPE algorithm, the battery parameter prediction device can obtain at least one first candidate battery parameter based on the preset battery parameter constraints and according to the first preset Gaussian mixture model and the second preset Gaussian mixture model through at least one iteration calculation.

[0196] It should be understood that the battery parameter prediction device can obtain a first candidate battery parameter through each iteration of calculation.

[0197] For example, for each iteration of calculation, the battery parameter prediction device can, based on preset battery parameter constraints, use the design hyperparameters corresponding to the first preset Gaussian mixture model, the test hyperparameters and process hyperparameters corresponding to the second preset Gaussian mixture model, which are the maximum ratio of the first preset Gaussian mixture model to the second preset Gaussian mixture model, as the first candidate battery parameters.

[0198] It should be understood that the battery parameter prediction device can use design hyperparameters as the first candidate design parameters among the first candidate battery parameters, test hyperparameters as the first candidate test parameters among the first candidate battery parameters, and process hyperparameters as the first candidate process parameters among the first candidate battery parameters.

[0199] Of course, the battery parameter prediction device can also generate at least one first candidate battery parameter by other means and according to a preset optimization algorithm based on preset battery parameter constraints.

[0200] Step S2032: Input the parameters of each first candidate battery into the rate prediction model to obtain the second predicted charging rate corresponding to each first candidate battery parameter.

[0201] In this step, the battery parameter prediction device inputs the parameters of each first candidate battery into the rate prediction model to obtain the second predicted charging rate corresponding to each first candidate battery parameter.

[0202] For example, the battery parameter prediction device can input each first candidate battery parameter into the rate prediction model to obtain the second predicted charging rate corresponding to each first candidate battery parameter, and store the second predicted charging rate corresponding to each first candidate battery parameter.

[0203] In another example, the battery parameter prediction device can sequentially input each first candidate battery parameter into the rate prediction model to obtain the second predicted charging rate corresponding to each first candidate battery parameter, and replace the historical second predicted charging rate with the latest second predicted charging rate.

[0204] In this application, at least one first candidate battery parameter is generated based on preset battery parameter constraints and a preset optimization algorithm. Further, each first candidate battery parameter is input into a rate prediction model to obtain a second predicted charging rate corresponding to each first candidate battery parameter. It is evident that in this embodiment, by combining preset battery parameter constraints and a preset optimization algorithm to determine the first candidate battery parameters, more suitable first candidate battery parameters can be generated. This allows for the determination of more suitable target candidate battery parameters based on the more suitable first candidate battery parameters and the corresponding second predicted charging rate, thereby facilitating the acquisition of more accurate ideal design parameters.

[0205] In some embodiments, Figure 4 is a schematic flowchart of a battery parameter prediction method provided in other embodiments of this application. Based on the above embodiments, the overall process of steps S203-S205 described above is illustrated in this application. As shown in Figure 4, the method of this application embodiment may include the following steps:

[0206] Step S401: Based on the preset battery parameter constraints, generate the latest first candidate battery parameters according to the preset optimization algorithm.

[0207] Step S402: Input the latest first candidate battery parameters into the rate prediction model to obtain the second predicted charging rate corresponding to the latest first candidate battery parameters.

[0208] Step S403: Based on the first predicted charging rate, the second predicted charging rate corresponding to the latest first candidate battery parameters, and the second predicted charging rate corresponding to the historical target candidate battery parameters, determine the latest target candidate battery parameters from the latest first candidate battery parameters and the historical target candidate battery parameters based on the first objective function.

[0209] Step S404: Determine whether the preset termination condition is met.

[0210] If the preset termination condition is met, proceed to step S405; if the preset termination condition is not met, return to step S401.

[0211] Step S405: Determine the ideal design parameters of the battery based on the latest target candidate battery parameters, and output the ideal design parameters of the battery.

[0212] It should be noted that steps S401-S405 above can be implemented using a preset parameter optimization model. This preset parameter optimization model can include, but is not limited to, a mathematical operations research optimization model, which can be solved using heuristic algorithms and exact algorithms.

[0213] Operations research typically employs mathematical methods to study optimization approaches and solutions for various systems, providing decision-makers with a scientific basis for decision-making. The primary research object of optimization methods is the management problems and production and operation activities of various organized systems. The purpose of optimization methods is to find the optimal solution for the rational use of human, material, and financial resources for the system under study, thereby maximizing and improving the system's efficiency and effectiveness, and ultimately achieving the system's optimal goal.

[0214] It should be understood that before executing step S401, after the battery parameter prediction device performs preliminary modeling of the preset parameter optimization model, it needs to store the decision variables, objective function, and preset battery parameter constraints of the preset parameter optimization model into the solution algorithm. Specifically, the settings may include, but are not limited to, at least one of the following: the reading path of the rate prediction model, the range of values ​​for the decision variables, the solution type and objective function type (e.g., maximization or minimization problem), and the expected value of the objective function in the preset parameter optimization model. The settings of the above optimization model content can be modified according to the specific optimization problem.

[0215] Furthermore, before running the solution algorithm, the battery parameter prediction device needs to pre-set a series of algorithm parameters to ensure the algorithm can function properly. These parameters may include, but are not limited to, at least one of the following: total number of iterations, algorithm type (e.g., random search algorithm, simulated annealing algorithm, or TPE algorithm). It should be noted that modifying these parameters can control the algorithm's solution speed and improve the optimization performance of the comparison algorithms.

[0216] As can be seen, the iterative optimization process in this application embodiment can yield more accurate ideal design parameters, thereby further improving the efficiency of battery research and development.

[0217] In some embodiments, Figure 5 is a flowchart illustrating a battery parameter prediction method provided in other embodiments of this application. Considering the existence of charging rate prediction boundaries in the rate prediction model, this application provides an exemplary description of the relevant content for determining the charging rate prediction boundaries. As shown in Figure 5, the method of this application embodiment may include the following steps:

[0218] Step S501: If the preset termination condition is not met, determine at least one second candidate battery parameter and the third predicted charging rate corresponding to each second candidate battery parameter based on the preset battery parameter constraint condition.

[0219] For example, the second candidate battery parameters involved in the embodiments of this application may include, but are not limited to, at least one of the second candidate design parameters, second candidate test parameters, and second candidate process parameters.

[0220] In one possible implementation, if the preset termination condition is not met, the battery parameter prediction device can determine at least one second candidate battery parameter based on the preset battery parameter constraints, and then determine the third predicted charging rate corresponding to each second candidate battery parameter according to the at least one second candidate battery parameter and the rate prediction model.

[0221] For example, the battery parameter prediction device can generate at least one second candidate battery parameter based on preset battery parameter constraints and according to a preset optimization algorithm, and input each second candidate battery parameter into the rate prediction model to obtain the third predicted charging rate corresponding to each second candidate battery parameter.

[0222] It should be noted that in the embodiments of this application, the feasible method of "generating at least one second candidate battery parameter according to a preset optimization algorithm based on preset battery parameter constraints, and inputting each second candidate battery parameter into the rate prediction model to obtain the third predicted charging rate corresponding to each second candidate battery parameter" can refer to the relevant content in the above embodiments regarding "generating at least one first candidate battery parameter according to a preset optimization algorithm based on preset battery parameter constraints, and inputting each first candidate battery parameter into the rate prediction model to obtain the second predicted charging rate corresponding to each first candidate battery parameter", which will not be repeated here.

[0223] In another possible implementation, the battery parameter prediction device can determine at least one second candidate battery parameter and a third predicted charging rate corresponding to each second candidate battery parameter by means of preset battery parameter constraints and preset battery parameter rate prediction model.

[0224] Of course, the battery parameter prediction device can also determine at least one second candidate battery parameter and the third predicted charging rate corresponding to each second candidate battery parameter in other ways based on preset battery parameter constraints.

[0225] Step S502: Based on the third predicted charging rate corresponding to each second candidate battery parameter, determine the candidate charging rate prediction boundary from each third predicted charging rate based on the second objective function.

[0226] In this step, the battery parameter prediction device can determine the candidate charging rate prediction boundary from each of the third predicted charging rates based on the third predicted charging rate corresponding to each of the second candidate battery parameters, using the second objective function.

[0227] The second objective function in this embodiment can be used to indicate the upper or lower boundary of the charging rate predicted by the rate prediction model. For example, the second objective function can be a function that minimizes the charging rate predicted by the rate prediction model.

[0228] It should be understood that the decision variables of the second objective function may include, but are not limited to, the variables corresponding to the input parameters of the rate prediction model. For example, the decision variables of the second objective function may include battery design parameter variables; of course, the decision variables of the second objective function may also include other variables, such as battery testing parameter variables, and / or process parameter variables.

[0229] For example, the second objective function in the embodiments of this application can be expressed as the following formula (2): f2=min P(x,y,z) Formula (2)

[0230] Here, f2 represents the second objective function.

[0231] Another example is that the second objective function can be a function that maximizes the charging rate predicted by the rate prediction model.

[0232] For example, the second objective function in the embodiments of this application can be expressed as the following formula (3): f2=max P(x,y,z) Formula (3)

[0233] It should be understood that for any second candidate battery parameter, when the battery parameter prediction device substitutes the second candidate design parameter into the design parameter variable x, the second candidate test parameter into the test parameter variable y, and the second candidate process parameter into the process parameter variable z, P(x,y,z) can represent the third predicted charging rate corresponding to the second candidate battery parameter.

[0234] Of course, the second objective function can also be expressed as other variations or equivalent formulas of the above formulas (2) or (3).

[0235] In one possible implementation, if the second objective function is a function that maximizes the charging rate predicted by the rate prediction model, the battery parameter prediction device can use the maximum value among the third predicted charging rates corresponding to each second candidate battery parameter as the upper boundary of the candidate charging rate prediction.

[0236] In this implementation, the battery parameter prediction device can determine the maximum predicted charging rate among the third predicted charging rates corresponding to each second candidate battery parameter based on the second objective function, and use the maximum predicted charging rate as the upper boundary of the candidate charging rate prediction, thereby achieving the purpose of exploring the upper boundary of the candidate charging rate prediction.

[0237] In another possible implementation, if the second objective function is a function that minimizes the charging rate predicted by the rate prediction model, the battery parameter prediction device can use the minimum value among the third predicted charging rates corresponding to each second candidate battery parameter as the lower boundary of the candidate charging rate prediction.

[0238] In this implementation, the battery parameter prediction device can achieve the purpose of exploring the lower boundary of the candidate charging rate prediction by using the minimum predicted charging rate among the third predicted charging rates corresponding to each second candidate battery parameter based on the second objective function, and taking the minimum predicted charging rate as the lower boundary of the candidate charging rate prediction.

[0239] It should be noted that the battery parameter prediction device can determine the upper boundary and lower boundary of the candidate charging rate prediction using the two types of second objective functions mentioned above.

[0240] Step S503: If the preset termination condition is met, the candidate charging rate prediction boundary is taken as the target charging rate prediction boundary.

[0241] In this step, under the condition that the preset termination condition is met, the battery parameter prediction device can use the candidate charging rate prediction boundary as the target charging rate prediction boundary of the rate prediction model. The candidate charging rate prediction boundary includes the upper boundary and / or the lower boundary of the candidate charging rate prediction; correspondingly, the target charging rate prediction boundary can include the upper boundary and / or the lower boundary of the target charging rate prediction.

[0242] In some embodiments, the battery parameter prediction device of this application can also output the target charging rate prediction boundary of the rate prediction model, so that researchers can know the charging rate prediction boundary of the rate prediction model.

[0243] In this embodiment, when a preset termination condition is not met, at least one second candidate battery parameter and a third predicted charging rate corresponding to each second candidate battery parameter are determined based on preset battery parameter constraints. Then, based on the third predicted charging rate corresponding to each second candidate battery parameter, a candidate charging rate prediction boundary is determined from each third predicted charging rate based on a second objective function. Further, when the preset termination condition is met, the candidate charging rate prediction boundary is taken as the target charging rate prediction boundary. Therefore, this embodiment, by determining the candidate charging rate prediction boundary from each third predicted charging rate based on the third predicted charging rate corresponding to each second candidate battery parameter and using a second objective function, can explore the charging rate prediction boundary of the rate prediction model. This facilitates researchers in understanding the rate prediction boundary of the rate prediction model, allowing them to select a rate prediction model more suitable for actual needs.

[0244] It should be understood that steps S501-S503 above can determine the target charging rate prediction boundary of the above rate prediction model through a preset rate prediction boundary model. The preset rate prediction boundary model can be, but is not limited to, a mathematical operations research optimization model.

[0245] The preset rate prediction boundary model in this application embodiment may include, but is not limited to, a preset rate prediction upper boundary model and / or a preset rate prediction lower boundary model. The preset rate prediction upper boundary model can be used to determine the upper bound of the charging rate prediction corresponding to the rate prediction model, and the preset rate prediction lower boundary model can be used to determine the lower bound of the charging rate prediction corresponding to the rate prediction model.

[0246] In some embodiments, the battery parameter prediction device of this application can also output the importance level of each feature parameter corresponding to the rate prediction model, so that researchers can know the feature parameters comprehensively considered by the rate prediction model, and thus focus on the feature parameters with higher importance during the battery parameter design process. The feature parameters corresponding to the rate prediction model may include, but are not limited to, at least one of design parameters, test parameters, and process parameters.

[0247] Figure 6 is a schematic diagram of the importance of each feature parameter corresponding to the ratio prediction model provided in some embodiments of this application. As shown in Figure 6, the horizontal axis is used to indicate the degree of importance, and the vertical axis is used to indicate different feature parameters. The schematic diagram of the degree of importance can display multiple feature parameters in order of increasing importance.

[0248] Figure 7 is a heat map of the correlation between the feature parameters corresponding to the ratio prediction model provided in some embodiments of this application. As shown in Figure 7, the horizontal and vertical axes are used to indicate different feature parameters. The greater the correlation between different feature parameters, the darker the color of the corresponding correlation heat map result.

[0249] In some embodiments, Figure 8 is a flowchart illustrating a training method for a multiplier prediction model provided in some embodiments of this application. Based on the above embodiments, this application provides an exemplary description of the training process of the multiplier prediction model involved in the above embodiments. As shown in Figure 8, the method of this application embodiment may include the following steps:

[0250] Step S801: Obtain the training sample set; wherein, each training sample in the training sample set may include: training battery parameters and corresponding training charging rate.

[0251] In the embodiments of this application, the training battery parameters in any training sample may include, but are not limited to: training design parameters, training testing parameters, and training process parameters.

[0252] In one possible implementation, the battery parameter prediction device can obtain a training sample set from a database or other devices, or it can also receive a training sample set input by researchers.

[0253] In another possible implementation, the battery parameter prediction device can acquire multiple sets of initial samples and obtain a training sample set by preprocessing the multiple sets of initial samples.

[0254] The initial sample in this application embodiment may include, but is not limited to, initial battery parameters and the corresponding initial charging rate. The initial battery parameters may include, but are not limited to, initial design parameters, initial test parameters, and initial process parameters.

[0255] Step S802: Train the initial prediction model based on the training sample set to obtain the multiplier prediction model.

[0256] In this step, the battery parameter prediction device can train the initial prediction model based on the training sample set to obtain the rate prediction model. The initial prediction model may include, but is not limited to, a machine learning model.

[0257] For example, the battery parameter prediction device can train an initial prediction model using machine learning algorithms such as Random Forest or XGBoost based on a training sample set to obtain a rate prediction model. Specifically, the battery parameter prediction device can set the prediction error of the machine learning model to the Mean Absolute Percentage Error (MAPE) and perform a regression fitting on the training sample set.

[0258] For example, during the prediction model tuning process, the battery parameter prediction device can use Bayesian optimization algorithms and other methods to continuously adjust the prediction model in order to find the optimal combination of hyperparameters so that the mean absolute percentage error (MAPE) of its prediction can be within 5%.

[0259] Random forest is a machine learning algorithm that incorporates multiple decision tree classifiers. It can train and predict samples using multiple trees. XGBoost is a distributed gradient boosting library that optimizes machine learning algorithms within the Gradient Boosting framework, enabling fast and accurate solutions to many data science problems.

[0260] Among them, the mean absolute percentage error (MAPE) refers to the average percentage of the absolute error between the predicted value and the actual value (i.e., the corresponding training charge rate), which is more suitable for problems where the target variable has a large difference in size.

[0261] Bayesian optimization is a sequential design strategy that performs global optimization of a black-box function without specifying any functional form. It is typically used to optimize complex evaluation functions.

[0262] Of course, the battery parameter prediction device can also train the initial prediction model using other methods based on the training sample set to obtain the rate prediction model.

[0263] In summary, in this embodiment, a training sample set is obtained; each training sample in the training sample set may include: training battery parameters and corresponding training charging rates; the training battery parameters include training design parameters, training test parameters, and training process parameters. Further, an initial prediction model is trained based on the training sample set to obtain a rate prediction model. It is evident that in this embodiment, the training of the initial prediction model considers information such as the battery's training design parameters, training test parameters, and training process parameters, taking into account a more comprehensive range of factors. Therefore, the rate prediction model trained in this embodiment has higher accuracy, enabling rapid and accurate prediction of the battery's predicted charging rate, thereby improving the efficiency of charging rate prediction.

[0264] In some embodiments, Figure 9 is a flowchart illustrating a method for obtaining a training sample set according to some embodiments of this application. Based on the above embodiments, this application provides an exemplary description of the relevant content of "obtaining the training sample set" in step S801. As shown in Figure 9, the method of this application embodiment may include the following steps:

[0265] Step S8011: Obtain multiple sets of initial samples, wherein the initial samples include initial battery parameters and corresponding initial charging rates; the initial battery parameters include initial design parameters, initial test parameters and initial process parameters.

[0266] In this step, the battery parameter prediction device can obtain multiple sets of initial samples from a database or other devices, or it can receive multiple sets of initial samples input by R&D personnel. It should be understood that the database can store a large amount of data resources generated during battery design and fabrication experiments and large-scale production.

[0267] For example, the battery parameter prediction device can acquire multiple initial design parameters and corresponding design identification information, and based on the multiple design identification information, acquire the initial test parameters, initial process parameters, and initial charging rates corresponding to the multiple design identification information respectively. Further, the battery parameter prediction device can merge the initial design parameters, initial test parameters, initial process parameters, and initial charging rates corresponding to the multiple design identification information into multiple sets of initial samples.

[0268] In this embodiment, the battery parameter prediction device can acquire multiple design identification information and corresponding initial design parameters. The design identification information can be used to uniquely identify the initial design parameters. For example, the design identification information may be, but is not limited to, the primary key information of a design identification package.

[0269] For example, a battery parameter prediction device can extract the primary key information of each design identifier package and the corresponding initial design parameters from a design database. For instance, the design database may include, but is not limited to, a Product Lifecycle Management (PLM) database; the primary key information of the design identifier package may include, but is not limited to, design group name or design group number, etc.

[0270] Furthermore, for each design identifier, the battery parameter prediction device can obtain the initial test parameters, initial process parameters, and initial charging rate corresponding to each design identifier.

[0271] For ease of understanding, the following embodiments of this application use the design identifier information as the primary key information of the design identifier package as an example for description.

[0272] For example, for each design identifier package primary key information, the battery parameter prediction device can use this design identifier package primary key information as the primary key to extract the corresponding initial test parameters, initial process parameters, and test results (i.e., initial charging rate) from the test database, process database, and test result database, respectively. For example, the test database may include, but is not limited to, the TRP test condition database, the process database may include, but is not limited to, the PES process database, and the test result database may include, but is not limited to, the TES test result database.

[0273] It should be noted that, considering data security, without disclosing data, the battery parameter prediction device can use encrypted information between data within the database to complete the data extraction and processing from the corresponding database, so as to form the modeling dataset required for the prediction model.

[0274] Furthermore, for each design identifier, the battery parameter prediction device can combine the initial design parameters, initial test parameters, initial process parameters, and initial charging rate corresponding to that design identifier into a set of initial samples.

[0275] For example, for each design identifier package primary key information, the battery parameter prediction device can use this design identifier package primary key information as the primary key to merge the corresponding initial design parameters, initial test parameters, initial process parameters, and initial charging rate into a set of initial samples. Among them, the initial design parameters, initial test parameters, and initial process parameters are characteristic attribute parameters (or simply characteristic parameters) corresponding to the initial charging rate.

[0276] Of course, battery parameter prediction devices can also obtain multiple sets of initial samples through other means.

[0277] Step S8012: Perform data preprocessing on multiple initial samples to obtain a training sample set.

[0278] In this step, the battery parameter prediction device can preprocess multiple sets of initial samples to obtain a training sample set. For example, data preprocessing may include, but is not limited to, at least one of the following: data filtering, data imputation, and data cleaning.

[0279] The data filtering process in this embodiment can be used to filter out feature data that is highly correlated with the charging rate from the initial battery parameters of multiple initial samples. This not only helps to improve the accuracy of the rate prediction model, but also reduces the computational cost in the modeling and use process.

[0280] The data imputation processing in this embodiment can be used to impute missing data in multiple initial samples, which can improve the data integrity of the training samples, thereby improving the data quality of the training samples and helping to improve the accuracy of the fold prediction model.

[0281] The data cleaning process in this embodiment can be used to clean abnormal data in multiple initial samples, which can improve the data accuracy of training samples, thereby improving the data quality of training samples and helping to improve the accuracy of the fold prediction model.

[0282] In one possible implementation, initial battery parameters and corresponding initial charging rates from multiple initial samples are filtered based on preset battery mechanism information to obtain first intermediate battery parameters and corresponding first intermediate charging rates from multiple sets of first intermediate samples whose influence on the charging rate is greater than a preset influence level. Further, the correlation between the first intermediate battery parameters and the charging rate in each of the multiple sets of first intermediate samples is determined, and based on the correlation of the first intermediate battery parameters in each of the multiple sets of first intermediate samples, the first intermediate battery parameters in each of the multiple sets of first intermediate samples are recursively deleted to obtain multiple sets of training samples whose correlation with the charging rate is greater than a preset correlation.

[0283] For example, the preset battery mechanism information in the embodiments of this application can be used to indicate the mechanism information for battery development; the initial battery parameters may include, but are not limited to: initial design parameters, initial test parameters and initial process parameters.

[0284] In this implementation, the battery parameter prediction device can determine the target feature attribute data that has a greater impact on the charging rate than the preset impact level based on the preset battery mechanism information. Then, based on the target feature attribute data, the device can filter the initial battery parameters and corresponding initial charging rates in multiple initial samples to obtain the first intermediate battery parameters and corresponding first intermediate charging rates in multiple first intermediate samples that have a greater impact on the charging rate than the preset impact level.

[0285] Furthermore, the battery parameter prediction device can determine the correlation between the first intermediate battery parameters and the charging rate in multiple sets of first intermediate samples according to a preset correlation algorithm. Further, the battery parameter prediction device can recursively delete first intermediate battery parameters from multiple sets of first intermediate samples by comparing the correlation of the first intermediate battery parameters in multiple sets of first intermediate samples with a preset correlation, thereby selecting multiple sets of first intermediate battery parameters and their corresponding first intermediate charging rates whose correlation with the charging rate is greater than the preset correlation, and using the selected multiple sets of first intermediate battery parameters and their corresponding first intermediate charging rates as training samples.

[0286] For example, the first intermediate design parameters selected by the battery parameter prediction device may include, but are not limited to, at least one of the following: upper main material, lower main material, platform voltage, and electrolyte.

[0287] For example, the first intermediate test parameter selected by the battery parameter prediction device may include, but is not limited to, at least one of the following: test temperature, test rate, and test time.

[0288] For example, the first intermediate process parameters screened by the battery parameter prediction device may include, but are not limited to, formation voltage and / or formation temperature.

[0289] In this embodiment of the application, by combining the basic knowledge of electrochemical mechanisms with statistical theory, multiple design parameters, test parameters, and process parameters that have the greatest impact on the charging rate are selected. This not only improves the accuracy of the rate prediction model, but also reduces the computational cost in the modeling and use process.

[0290] In another possible implementation, based on the synchronization fault information of the database used to store battery information, the initial battery parameters and corresponding initial charging rates in multiple initial samples are subjected to a first data imputation process to obtain the second intermediate battery parameters and corresponding second intermediate charging rates in multiple second intermediate samples; further, the second intermediate battery parameters and corresponding second intermediate charging rates in multiple second intermediate samples are subjected to a second data imputation process according to a preset imputation method to obtain multiple training samples.

[0291] Since the data stored in the database is usually imported by researchers after recording experimental data, data loss can sometimes occur due to human error. In addition, it's also possible that a write error occurs during the data import and merging process, causing data to fail to synchronize in a timely manner and resulting in data loss.

[0292] In this implementation, the battery parameter prediction device can locate the synchronization fault information of the database used to store battery information, and perform first data filling processing on the initial battery parameters and corresponding initial charging rates in multiple initial samples according to the located synchronization fault information of the database, so as to obtain the second intermediate battery parameters and corresponding second intermediate charging rates in multiple second intermediate samples.

[0293] Furthermore, the battery parameter prediction device can perform second data imputation processing on the second intermediate battery parameters and corresponding second intermediate charging rates in multiple sets of second intermediate samples according to a preset imputation method, to obtain multiple sets of imputed second intermediate battery parameters and corresponding second intermediate charging rates, and use the imputed multiple sets of second intermediate battery parameters and corresponding second intermediate charging rates as training samples. The preset imputation method may include, but is not limited to, at least one of the following: an imputation method based on preset battery mechanism information, a preset interpolation imputation method, or a preset regression fitting imputation method.

[0294] In this embodiment of the application, by combining the filling method of synchronous fault information in the database with the preset filling method, the data integrity of the training samples can be improved, thereby improving the data quality of the training samples and helping to improve the accuracy of the ratio prediction model.

[0295] Considering that some of the data generated during battery experiments is measured using instruments and then manually recorded, there may be measurement errors in the instruments, and / or errors may occur due to the manual data recording method.

[0296] In another possible implementation, the battery parameter prediction device can remove data that exceeds the corresponding preset parameter boundaries from the initial battery parameters and corresponding initial charging rates in multiple initial samples based on preset battery mechanism information, thereby obtaining the third intermediate battery parameters and third intermediate charging rates in multiple third intermediate samples.

[0297] Furthermore, the battery parameter prediction device can divide the third intermediate battery parameters and the corresponding third intermediate charging rate in multiple sets of third intermediate samples into multiple sets of intermediate parameters according to different test conditions; wherein, each set of intermediate parameters includes multiple third intermediate battery parameters and the corresponding third intermediate charging rate.

[0298] Furthermore, the battery parameter prediction device can calculate the mean absolute percentage error within multiple sets of intermediate parameters, and remove data with a mean absolute percentage error within the set that is greater than a preset error, thus obtaining multiple sets of training samples.

[0299] For example, for each set of intermediate parameters, the battery parameter prediction device can remove data whose average absolute percentage error in the same set of intermediate parameters is greater than a preset error (e.g., 10%), thus obtaining the remaining third intermediate battery parameters and corresponding third intermediate charging rates in that set of intermediate parameters. Further, the battery parameter prediction device can use the remaining third intermediate battery parameters and corresponding third intermediate charging rates in each set of intermediate parameters as training samples.

[0300] In this application, by combining preset parameter boundaries and mean absolute percentage error for data cleaning, the accuracy of training sample data can be improved, thereby improving the data quality of training samples and thus improving the accuracy of the ratio prediction model.

[0301] It should be noted that the battery parameter prediction device can perform the above-mentioned data preprocessing on multiple sets of initial samples to obtain a training sample set.

[0302] Of course, battery parameter prediction devices can also preprocess multiple initial samples in other ways to obtain a training sample set.

[0303] In this embodiment, multiple sets of initial samples are obtained, including initial battery parameters and corresponding initial charging rates. The initial battery parameters include initial design parameters, initial test parameters, and initial process parameters. Further, data preprocessing is performed on these multiple sets of initial samples to obtain a training sample set. Therefore, in this embodiment, by preprocessing the obtained multiple sets of initial samples, the data quality of the training samples can be improved, which is beneficial to improving the accuracy of the rate prediction model and thus improving its prediction precision.

[0304] In some embodiments, based on the above embodiments, this application combines the rate prediction model and the mathematical operations research optimization model to describe the overall battery parameter prediction model of this application. Figure 10 is a schematic diagram of the structure of the battery parameter prediction model provided in this application embodiment. As shown in Figure 10, the battery parameter prediction model may include a rate prediction model and a mathematical operations research optimization model; the mathematical operations research optimization model may include, but is not limited to, the above-mentioned preset parameter optimization model and / or preset rate prediction boundary model; wherein, the preset rate prediction boundary model may include, but is not limited to, a preset rate prediction upper boundary model and / or a preset rate prediction lower boundary model.

[0305] 1) Ratio Prediction Model

[0306] Battery parameter prediction equipment can obtain initial design parameters, initial test parameters, initial charge rate, and initial process parameters from the design database, test condition database, test result database, and process database, respectively.

[0307] Furthermore, the battery parameter prediction device can obtain a modeling dataset by performing data imputation and data cleaning on the initial battery parameters.

[0308] It should be understood that the battery parameter prediction device can divide the modeling dataset into a training dataset and a test dataset according to a preset ratio (e.g., 8:2).

[0309] Furthermore, the battery parameter prediction device can perform data modeling based on the modeling dataset and obtain a rate prediction model through machine learning.

[0310] 2) Mathematical Operations Research Optimization Model

[0311] The battery parameter prediction device allows setting decision variables, objective functions, and battery parameter constraints for a mathematical operations research optimization model. It should be noted that the mathematical operations research optimization model can include multiple models, each with a different objective function.

[0312] Furthermore, the battery parameter prediction device can perform mathematical modeling and algorithm parameter setting to obtain a mathematical operations research optimization model.

[0313] In one possible implementation, the rate prediction model and the mathematical operations optimization model in this application embodiment can be deployed separately, so that researchers can obtain the predicted charging rate by inputting battery parameters into the rate prediction model, and obtain the ideal design parameters corresponding to the charging rate by inputting the required charging rate into the mathematical operations optimization model.

[0314] In another possible implementation, the rate prediction model and the mathematical operations research optimization model in this embodiment can be combined into a battery parameter prediction model. This allows researchers to input battery parameters into the battery parameter prediction model, which can not only quickly output the predicted charging rate, but also output the ideal design parameters that are closest to the predicted charging rate. Furthermore, it can also output charging rate prediction boundary information, etc., from the battery parameter prediction model.

[0315] In some embodiments, based on the above embodiments, the following embodiments of this application provide an exemplary description of the prediction accuracy of the rate prediction model, the preset charging rate prediction boundary, and the optimization of design parameters.

[0316] 1) Results of the prediction accuracy of the ratio prediction model

[0317] Figure 11 is a schematic diagram of the prediction accuracy of the ratio prediction model provided in the embodiment of this application. As shown in Figure 11, in the embodiment of this application, the fitting error MAPE of the ratio prediction model based on the training dataset is equal to 0.0042.

[0318] Figure 12 is a schematic diagram of the prediction accuracy of the ratio prediction model provided in the embodiment of this application. As shown in Figure 12, in the embodiment of this application, the fitting error MAPE of the ratio prediction model based on the test dataset is equal to 0.011.

[0319] 2) Display of results for preset charging rate prediction boundaries

[0320] The battery parameter prediction device can determine the charging rate prediction boundary of the above-mentioned rate prediction model through a preset rate prediction boundary model. The preset rate prediction boundary model can include a preset upper rate prediction boundary model and a preset lower rate prediction boundary model.

[0321] For example, the battery parameter prediction device can use a random search algorithm, a simulated annealing algorithm, or a TPE algorithm to solve the preset rate prediction boundary model. The solution results of different algorithms can be referred to Table 1 (Table 1 is a schematic table of results corresponding to different heuristic algorithms).

[0322] Table 1 shows the results for different heuristic algorithms.

[0323] As shown in Table 1, after using the three algorithm strategies, the TPE algorithm can solve for the maximum charging rate prediction range of the rate prediction model. Therefore, the final prediction interval of the rate prediction model can be [0.6936, 1.0144].

[0324] 3) Demonstration of the prediction accuracy of the ratio prediction model

[0325] Battery parameter prediction equipment can determine the ideal design parameters corresponding to achieving the target predicted charging rate by using a preset parameter optimization model.

[0326] For example, the battery parameter prediction device can use a random search algorithm, a simulated annealing algorithm, or a TPE algorithm to solve the preset parameter optimization model. The solution results of different algorithms can be referred to Tables 2-4 below.

[0327] Table 2 shows the results of different heuristic algorithms when the target predicted charging rate is close to 0.8.

[0328] As shown in Table 2, after solving the optimization model with preset parameters using three different algorithmic strategies, the TPE algorithm can find the solution with the smallest error from the optimal solution, but it consumes more time. In contrast, the simulated annealing algorithm can find the best solution to date in a shorter time, but this solution has the largest error from the optimal solution. The random search algorithm's optimization performance is between that of the TPE algorithm and the simulated annealing algorithm, and it can find even better solutions in the later stages of iteration.

[0329] Figure 13 is an iterative diagram of the objective function value of the random search algorithm provided in the embodiment of this application. As shown in Figure 13, the horizontal axis is used to indicate the number of iterations, and the vertical axis is used to indicate the optimal objective function value so far. Figure 13 can be used to indicate the change of the optimal objective function value of the random search algorithm as the number of iterations increases.

[0330] Figure 14 is a schematic diagram of the frequency difference of the objective function value of the random search algorithm provided in the embodiment of this application. As shown in Figure 14, the horizontal axis is used to indicate the absolute value of the difference between the objective function value and the performance target in each iteration, and the vertical axis is used to indicate the frequency of occurrence. Figure 14 can be used to indicate the frequency of occurrence of the absolute value of the difference between the objective function value and the performance target in each iteration of the random search algorithm.

[0331] Figure 15 is an iterative diagram of the objective function value of the simulated annealing algorithm provided in the embodiment of this application. Figure 15 can be used to indicate the change of the optimal objective function value of the simulated annealing algorithm as the number of iterations increases.

[0332] Figure 16 is a schematic diagram of the frequency difference histogram of the objective function value of the simulated annealing algorithm provided in the embodiment of this application. Figure 16 can be used to indicate the frequency of the absolute value of the difference between the objective function value and the performance target in each iteration of the simulated annealing algorithm.

[0333] Figure 17 is an iterative diagram of the objective function value of the TPE algorithm provided in the embodiment of this application. Figure 17 can be used to indicate the change of the optimal objective function value of the TPE algorithm as the number of iterations increases.

[0334] Figure 18 is a schematic diagram of the frequency difference histogram of the objective function value of the TPE algorithm provided in the embodiment of this application. Figure 18 can be used to indicate the frequency of the absolute value of the difference between the objective function value and the performance target in each iteration of the TPE algorithm.

[0335] Table 3 shows the results of different heuristic algorithms when the target predicted charging rate is close to 0.9.

[0336] As shown in Table 3, after solving the optimization model with preset parameters using three different algorithmic strategies, the random search algorithm and the TPE algorithm can obtain the solution with the smallest error from the optimal solution. Among them, the random search algorithm can obtain its better result in a shorter time. In contrast, although the simulated annealing algorithm can obtain the best solution to date in a shorter time, this solution has the largest error from the optimal solution.

[0337] Figure 19 is a second schematic diagram of the iteration of the objective function value of the random search algorithm provided in the embodiment of this application, and Figure 20 is a second schematic diagram of the difference frequency histogram of the objective function value of the random search algorithm provided in the embodiment of this application.

[0338] Figure 21 is a second schematic diagram of the iteration of the objective function value of the simulated annealing algorithm provided in the embodiment of this application, and Figure 22 is a second schematic diagram of the difference frequency histogram of the objective function value of the simulated annealing algorithm provided in the embodiment of this application.

[0339] Figure 23 is a second schematic diagram of the iteration of the objective function value of the TPE algorithm provided in the embodiment of this application, and Figure 24 is a second schematic diagram of the difference frequency histogram of the objective function value of the TPE algorithm provided in the embodiment of this application.

[0340] Table 4 shows the results of different heuristic algorithms when the target predicted charging rate is close to 1.0.

[0341] As shown in Table 4, after solving the optimization model with preset parameters using three different algorithmic strategies, the TPE algorithm, while consuming relatively more time, can find the solution with the smallest error from the optimal solution. In contrast, the random search algorithm can find a relatively good solution to date in a shorter time. The simulated annealing algorithm has the worst optimization performance; although it finds the best solution to date in the shortest time (24th iteration), it cannot escape the best solution to date in the subsequent 1976 iterations.

[0342] Figure 25 is a schematic diagram of the iteration of the objective function value of the random search algorithm provided in the embodiment of this application, and Figure 26 is a schematic diagram of the difference frequency histogram of the objective function value of the random search algorithm provided in the embodiment of this application.

[0343] Figure 27 is a schematic diagram of the iteration of the objective function value of the simulated annealing algorithm provided in the embodiment of this application, and Figure 28 is a schematic diagram of the difference frequency histogram of the objective function value of the simulated annealing algorithm provided in the embodiment of this application.

[0344] Figure 29 is a schematic diagram of the iteration of the objective function value of the TPE algorithm provided in the embodiment of this application, and Figure 30 is a schematic diagram of the difference frequency histogram of the objective function value of the TPE algorithm provided in the embodiment of this application.

[0345] In summary, the preset parameter optimization model in this application utilizes operations research principles, designing and developing multiple efficient heuristic algorithms to explore the charging rate prediction boundary of the rate prediction model and optimize the design parameters to meet a certain charging rate. Comparative experiments between different algorithms demonstrate the iterative behavior of the algorithms under different strategies, and show that simulated annealing and Bayesian optimization algorithms can achieve the optimization goal faster with fewer iterations.

[0346] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0347] Based on the same inventive concept, this application also provides a battery parameter prediction device for implementing the battery parameter prediction method described above. The solution provided by this device is similar to the implementation described in the above method; therefore, the specific limitations in one or more battery parameter prediction device embodiments provided below can be found in the limitations of the battery parameter prediction method described above, and will not be repeated here.

[0348] In some embodiments, FIG31 is a schematic diagram of the structure of a battery parameter prediction device provided in some embodiments of this application. The battery parameter prediction device provided in the embodiments of this application can be applied to a battery parameter prediction device. As shown in FIG31, the battery parameter prediction device in the embodiments of this application may include: a first acquisition module 3101, a first determination module 3102, a second determination module 3103, a third determination module 3104, and a fourth determination module 3105.

[0349] The first acquisition module 3101 is used for battery parameters, which include design parameters, test parameters and process parameters.

[0350] The first determining module 3102 is used to input battery parameters into the rate prediction model and determine the first predicted charging rate of the battery based on the output of the rate prediction model.

[0351] The second determining module 3103 is used to determine at least one first candidate battery parameter and a second predicted charging rate corresponding to each first candidate battery parameter based on preset battery parameter constraints when the preset termination condition is not met; wherein the preset battery parameter constraints include preset design parameter constraints, preset test parameter constraints and preset process parameter constraints.

[0352] The third determining module 3104 is used to determine the target candidate battery parameters from each first candidate battery parameter based on the first predicted charging rate and the second predicted charging rate corresponding to each first candidate battery parameter.

[0353] The fourth determining module 3105 is used to take the target candidate battery parameters as the ideal design parameters of the battery when the preset termination conditions are met.

[0354] In some embodiments, the third determining module 3104 is specifically used for:

[0355] Based on the first predicted charging rate and the second predicted charging rate corresponding to each first candidate battery parameter, the second predicted charging rate when the function value of the first objective function is minimized is determined based on the first objective function.

[0356] The parameters of the first candidate battery corresponding to the second predicted charging rate are used as the target candidate battery parameters.

[0357] In some embodiments, the first objective function is a function that minimizes the absolute value of the difference between the first predicted charging rate and the charging rate predicted by the rate prediction model.

[0358] In some embodiments, the second determining module 3103 includes:

[0359] The generation unit is used to generate at least one first candidate battery parameter based on preset battery parameter constraints and according to a preset optimization algorithm.

[0360] The determining unit is used to input the parameters of each first candidate battery into the rate prediction model to obtain the second predicted charging rate corresponding to each first candidate battery parameter.

[0361] In some embodiments, the generating unit is specifically used for:

[0362] Based on preset battery parameter constraints, at least one first candidate battery parameter is randomly generated.

[0363] In some embodiments, the generating unit is specifically used for:

[0364] Based on preset battery parameter constraints, at least one first candidate battery parameter is obtained through at least one iterative calculation according to the first preset Gaussian mixture model and the second preset Gaussian mixture model.

[0365] In some embodiments, the generating unit is specifically used for:

[0366] For each iteration of the calculation, based on the preset battery parameter constraints, when the ratio of the first preset Gaussian mixture model to the second preset Gaussian mixture model is maximized, the design hyperparameters corresponding to the first preset Gaussian mixture model, as well as the test hyperparameters and process hyperparameters corresponding to the second preset Gaussian mixture model, are used as the first candidate battery parameters.

[0367] In some embodiments, the battery parameter prediction device of this application may further include:

[0368] The fifth determining module is used to determine at least one second candidate battery parameter and a third predicted charging rate corresponding to each second candidate battery parameter based on preset battery parameter constraints when the preset termination condition is not met.

[0369] The sixth determining module is used to determine the candidate charging rate prediction boundary from each third predicted charging rate based on the third predicted charging rate corresponding to each second candidate battery parameter and the second objective function.

[0370] The seventh determination module is used to use the candidate charging rate prediction boundary as the target charging rate prediction boundary when the preset termination conditions are met.

[0371] In some embodiments, if the second objective function is a function that maximizes the charging rate predicted by the rate prediction model, the sixth determining module is specifically used for:

[0372] The maximum value among the third predicted charging rates corresponding to each of the second candidate battery parameters is taken as the upper boundary of the candidate charging rate prediction.

[0373] In some embodiments, if the second objective function is a function that minimizes the charging rate predicted by the rate prediction model, the sixth determining module is specifically used for:

[0374] The minimum value among the third predicted charging rates corresponding to each of the second candidate battery parameters is taken as the lower boundary of the candidate charging rate prediction.

[0375] In some embodiments, the fifth determining module is specifically used for:

[0376] Based on preset battery parameter constraints, at least one second candidate battery parameter is generated according to a preset optimization algorithm.

[0377] The parameters of each second candidate battery are input into the rate prediction model to obtain the third predicted charging rate corresponding to each second candidate battery parameter.

[0378] In some embodiments, the battery parameter prediction device of this application may further include:

[0379] The second acquisition module is used to acquire a training sample set; wherein, each training sample in the training sample set includes: training battery parameters and corresponding training charging rates; the training battery parameters include training design parameters, training test parameters and training process parameters;

[0380] The training module is used to train the initial prediction model based on the training sample set to obtain the ratio prediction model.

[0381] In some embodiments, the second acquisition module includes:

[0382] The acquisition unit is used to acquire multiple sets of initial samples, wherein the initial samples include initial battery parameters and corresponding initial charging rates; the initial battery parameters include initial design parameters, initial test parameters and initial process parameters;

[0383] The processing unit is used to preprocess multiple sets of initial samples to obtain a training sample set.

[0384] In some embodiments, the obtaining unit is specifically used for:

[0385] Obtain multiple initial design parameters and their corresponding design identification information;

[0386] Based on multiple design identifiers, obtain the initial test parameters, initial process parameters, and initial charging rate corresponding to each design identifier;

[0387] The initial design parameters, initial test parameters, initial process parameters, and initial charging rates corresponding to multiple design identification information are merged into multiple initial samples.

[0388] In some embodiments, the processing unit is specifically used for:

[0389] Based on the preset battery mechanism information, the initial battery parameters and corresponding initial charging rates in multiple initial samples are screened to obtain the first intermediate battery parameters and corresponding first intermediate charging rates in multiple first intermediate samples whose influence on the charging rate is greater than the preset influence level.

[0390] The correlation between the first intermediate battery parameters and the charging rate in multiple sets of first intermediate samples was determined.

[0391] Based on the correlation of the first intermediate battery parameters in multiple sets of first intermediate samples, the first intermediate battery parameters in multiple sets of first intermediate samples are recursively deleted to obtain multiple sets of training samples with a correlation greater than the preset correlation with the charging rate.

[0392] In some embodiments, the processing unit is specifically used for:

[0393] Based on the synchronization fault information of the database used to store battery information, the initial battery parameters and corresponding initial charging rates in multiple initial samples are subjected to the first data filling process to obtain the second intermediate battery parameters and corresponding second intermediate charging rates in multiple second intermediate samples.

[0394] According to the preset filling method, the second intermediate battery parameters and the corresponding second intermediate charging rate in multiple sets of second intermediate samples are subjected to second data filling processing to obtain multiple sets of training samples.

[0395] In some embodiments, the processing unit is specifically used for:

[0396] Based on the preset battery mechanism information, the data that exceed the corresponding preset parameter boundary in the initial battery parameters and corresponding initial charging rates of multiple initial samples are removed to obtain the third intermediate battery parameters and third intermediate charging rates in multiple third intermediate samples.

[0397] Based on different test conditions, the third intermediate battery parameters and corresponding third intermediate charging rates in multiple sets of third intermediate samples are divided into multiple sets of intermediate parameters; each set of intermediate parameters includes multiple third intermediate battery parameters and corresponding third intermediate charging rates.

[0398] Calculate the mean absolute percentage error within each of the multiple intermediate parameter sets, and remove data whose mean absolute percentage error within each set is greater than the preset error to obtain multiple training samples.

[0399] The battery parameter prediction device provided in this application embodiment can be used to execute the technical solutions in the above-described battery parameter prediction method embodiments of this application. Its implementation principle and technical effect are similar, and will not be repeated here.

[0400] Each module in the aforementioned battery parameter prediction device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of the battery parameter prediction device in hardware form or independent of it, or they can be stored in the memory of the battery parameter prediction device in software form, so that the processor can call and execute the corresponding operations of each module.

[0401] In some embodiments, FIG32 is a schematic diagram of the structure of a battery parameter prediction device provided in some embodiments of this application. As shown in FIG32, the battery parameter prediction device provided in the embodiments of this application may include a processor, a memory, and a communication interface connected via a system bus. The processor of the battery parameter prediction device provides computing and control capabilities. The memory of the battery parameter prediction device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface of the battery parameter prediction device is used for wired or wireless communication with external devices. When the computer program is executed by the processor, it implements the technical solutions in the above-described embodiments of the battery parameter prediction method of this application. The implementation principle and technical effects are similar and will not be repeated here.

[0402] Those skilled in the art will understand that the structure shown in Figure 32 is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the battery parameter prediction device to which the present application is applied. A specific battery parameter prediction device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0403] In some embodiments, a battery parameter prediction device is also provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the technical solution in the above-described battery parameter prediction method embodiments of this application. The implementation principle and technical effect are similar, and will not be repeated here.

[0404] In some embodiments, a computer-readable storage medium is also provided, on which a computer program is stored. When the computer program is executed by a processor, it implements the technical solution in the above-described battery parameter prediction method embodiments of this application. Its implementation principle and technical effect are similar, and will not be repeated here.

[0405] In some embodiments, a computer program product is also provided, including a computer program that, when executed by a processor, implements the technical solutions in the battery parameter prediction method embodiments described above. The implementation principle and technical effects are similar and will not be repeated here.

[0406] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, feature databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The processors involved in the embodiments provided in this application can be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based feature data processing logic devices, etc., and are not limited thereto.

[0407] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and not to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. These modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application, and they should all be covered within the scope of the claims and specification of this application. In particular, as long as there is no structural conflict, the various technical features mentioned in the embodiments can be combined in any way. This application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A battery parameter prediction method, wherein, The method includes: Obtain the battery parameters, which include design parameters, test parameters, and process parameters; The battery parameters are input into the rate prediction model, and the first predicted charging rate of the battery is determined based on the output of the rate prediction model. If the preset termination condition is not met, at least one first candidate battery parameter and a second predicted charging rate corresponding to each first candidate battery parameter are determined based on preset battery parameter constraints; wherein, the preset battery parameter constraints include preset design parameter constraints, preset test parameter constraints and preset process parameter constraints. Based on the first predicted charging rate and the second predicted charging rate corresponding to each of the first candidate battery parameters, the target candidate battery parameters are determined from each of the first candidate battery parameters. If the preset termination condition is met, the ideal design parameters of the battery are determined based on the target candidate battery parameters.

2. The method of claim 1, wherein, The step of determining target candidate battery parameters from each of the first candidate battery parameters based on the first predicted charging rate and the second predicted charging rate corresponding to each of the first candidate battery parameters includes: Based on the first predicted charging rate and the second predicted charging rate corresponding to each of the first candidate battery parameters, the second predicted charging rate when the function value of the first objective function is minimized is determined based on the first objective function. The first candidate battery parameter corresponding to the second predicted charging rate is used as the target candidate battery parameter.

3. The method of claim 2, wherein, The first objective function is a function that takes the minimum value of the absolute value of the difference between the first predicted charging rate and the charging rate predicted by the rate prediction model.

4. The method of any one of claims 1-3, wherein, The step of determining at least one first candidate battery parameter and the second predicted charging rate corresponding to each first candidate battery parameter based on preset battery parameter constraints includes: Based on preset battery parameter constraints, at least one first candidate battery parameter is generated according to a preset optimization algorithm; Each of the first candidate battery parameters is input into the rate prediction model to obtain the second predicted charging rate corresponding to each of the first candidate battery parameters.

5. The method of claim 4, wherein, The step of generating at least one first candidate battery parameter based on preset battery parameter constraints and a preset optimization algorithm includes: Based on preset battery parameter constraints, at least one of the first candidate battery parameters is randomly generated; or, Based on preset battery parameter constraints, at least one first candidate battery parameter is obtained through at least one iterative calculation according to the first preset Gaussian mixture model and the second preset Gaussian mixture model.

6. The method of claim 5, wherein, Based on preset battery parameter constraints, and according to a first preset Gaussian mixture model and a second preset Gaussian mixture model, at least one first candidate battery parameter is obtained through at least one iteration calculation, including: For each iteration of the calculation, based on the preset battery parameter constraints, when the ratio of the first preset Gaussian mixture model to the second preset Gaussian mixture model is maximized, the design hyperparameters corresponding to the first preset Gaussian mixture model, as well as the test hyperparameters and process hyperparameters corresponding to the second preset Gaussian mixture model, are used as the first candidate battery parameters.

7. The method of any one of claims 1-6, wherein, The method further includes: If the preset termination condition is not met, at least one second candidate battery parameter and a third predicted charging rate corresponding to each second candidate battery parameter are determined based on the preset battery parameter constraint condition. Based on the third predicted charging rate corresponding to each of the second candidate battery parameters, the candidate charging rate prediction boundary is determined from each of the third predicted charging rates based on the second objective function. If the preset termination condition is met, the candidate charging rate prediction boundary is taken as the target charging rate prediction boundary.

8. The method of claim 7, wherein, If the second objective function is a function that maximizes the charging rate predicted by the rate prediction model, the step of determining the candidate charging rate prediction boundary from each of the third predicted charging rates based on the second objective function according to the third predicted charging rates corresponding to each of the second candidate battery parameters includes: The maximum value among the third predicted charging rates corresponding to each of the second candidate battery parameters is taken as the upper boundary of the candidate charging rate prediction.

9. The method of claim 7, wherein, If the second objective function is a function that minimizes the charging rate predicted by the rate prediction model, the step of determining the candidate charging rate prediction boundary from each of the third predicted charging rates based on the second objective function according to the third predicted charging rates corresponding to each of the second candidate battery parameters includes: The minimum value among the third predicted charging rates corresponding to each of the second candidate battery parameters is taken as the lower boundary of the candidate charging rate prediction.

10. The method of any one of claims 7-9, wherein, The step of determining at least one second candidate battery parameter and a third predicted charging rate corresponding to each second candidate battery parameter based on the preset battery parameter constraints includes: Based on the preset battery parameter constraints, at least one second candidate battery parameter is generated according to the preset optimization algorithm. Each of the second candidate battery parameters is input into the rate prediction model to obtain the third predicted charging rate corresponding to each of the second candidate battery parameters.

11. The method of any one of claims 1-10, wherein, The method further includes: Obtain a training sample set; wherein each training sample in the training sample set includes: training battery parameters and corresponding training charging rates; the training battery parameters include training design parameters, training test parameters and training process parameters; The initial prediction model is trained based on the training sample set to obtain the ratio prediction model.

12. The method of claim 11, wherein, The acquisition of the training sample set includes: Multiple sets of initial samples are obtained, wherein the initial samples include initial battery parameters and corresponding initial charging rates; the initial battery parameters include initial design parameters, initial test parameters and initial process parameters; The training sample set is obtained by preprocessing the multiple initial samples.

13. The method of claim 12, wherein, The process of obtaining multiple sets of initial samples includes: Obtain multiple initial design parameters and their corresponding design identification information; Based on the multiple design identification information, obtain the initial test parameters, initial process parameters and initial charging rate corresponding to the multiple design identification information respectively; The initial design parameters, initial test parameters, initial process parameters, and initial charging rates corresponding to the multiple design identification information are merged into multiple initial samples.

14. The method of claim 12 or 13, wherein, The step of preprocessing the multiple sets of initial samples to obtain the training sample set includes: Based on the preset battery mechanism information, the initial battery parameters and corresponding initial charging rates in multiple sets of initial samples are screened to obtain the first intermediate battery parameters and corresponding first intermediate charging rates in multiple sets of first intermediate samples whose influence on the charging rate is greater than the preset influence level. The correlation between the first intermediate battery parameters and the charging rate in multiple sets of the first intermediate samples was determined respectively; Based on the correlation of the first intermediate battery parameters in multiple sets of the first intermediate samples, the first intermediate battery parameters in multiple sets of the first intermediate samples are recursively deleted to obtain multiple sets of training samples with a correlation greater than the preset correlation with the charging rate.

15. The method of any one of claims 12-14, wherein, The step of preprocessing the multiple sets of initial samples to obtain the training sample set includes: Based on the synchronization fault information of the database used to store battery information, the initial battery parameters and corresponding initial charging rates in multiple sets of initial samples are subjected to first data filling processing to obtain the second intermediate battery parameters and corresponding second intermediate charging rates in multiple sets of second intermediate samples. According to the preset filling method, the second intermediate battery parameters and the corresponding second intermediate charging rate in multiple sets of the second intermediate samples are subjected to second data filling processing to obtain multiple sets of training samples.

16. The method of any one of claims 12-15, wherein, The step of preprocessing the multiple sets of initial samples to obtain the training sample set includes: Based on the preset battery mechanism information, the data that exceed the corresponding preset parameter boundary in the initial battery parameters and corresponding initial charging rates of the multiple sets of initial samples are removed to obtain the third intermediate battery parameters and third intermediate charging rates in the multiple sets of third intermediate samples. Based on different test conditions, the third intermediate battery parameters and corresponding third intermediate charge rates in the multiple sets of third intermediate samples are divided into multiple sets of intermediate parameters; wherein, each set of intermediate parameters includes multiple sets of third intermediate battery parameters. Battery parameters and the corresponding third intermediate charging rate; Calculate the mean absolute percentage error within each of the multiple sets of intermediate parameters, and remove data whose mean absolute percentage error within each set is greater than a preset error to obtain multiple sets of training samples.

17. A battery parameter prediction device, wherein, The device includes: The module is used to acquire battery parameters, which include design parameters, test parameters, and process parameters. The first determining module is used to input the battery parameters into the rate prediction model and determine the first predicted charging rate of the battery based on the output of the rate prediction model. The second determining module is used to determine at least one first candidate battery parameter and a second predicted charging rate corresponding to each first candidate battery parameter based on preset battery parameter constraints when the preset termination condition is not met; wherein the preset battery parameter constraints include preset design parameter constraints, preset test parameter constraints and preset process parameter constraints. The third determining module is used to determine the target candidate battery parameters from each of the first candidate battery parameters based on the first predicted charging rate and the second predicted charging rate corresponding to each of the first candidate battery parameters. The fourth determining module is used to use the target candidate battery parameters as the ideal design parameters of the battery when the preset termination condition is met.

18. A battery parameter prediction device, wherein, The method includes a memory and a processor, the memory storing a computer program, wherein the processor executes the computer program to implement the steps of the method according to any one of claims 1-16.

19. A computer readable storage medium having stored thereon a computer program, wherein, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1-16.

20. A computer program product comprising a computer program, wherein, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1-16.

Images