CN119959764A

CN119959764A - Battery health status prediction method and device

Info

Publication number: CN119959764A
Application number: CN202311491388.5A
Authority: CN
Inventors: 郑沐穹
Original assignee: Xiaomi Automobile Technology Co Ltd
Current assignee: Xiaomi Automobile Technology Co Ltd
Priority date: 2023-11-09
Filing date: 2023-11-09
Publication date: 2025-05-09

Abstract

The present disclosure provides a method and device for predicting the health status of a battery, which relates to the field of simulation operations. The specific steps are: obtaining a first training data set, training a battery status prediction model according to the first training data set to determine a first parameter set, wherein the first training data set is the operating condition data of the battery during operation; obtaining a second training data set, training a battery status prediction model according to the second training data set to determine a second parameter set, wherein the second training data set is data related to the health status of the battery; configuring the first parameter set and the second parameter set into the battery status prediction model; inputting the real-time parameters of the battery into the battery status prediction model for inference operations to determine the battery health status parameters. The present disclosure trains the parameters in the battery status prediction model by two training data sets, thereby realizing the prediction of the battery health status parameters, avoiding the problem of insufficient prediction accuracy, and improving the accuracy of the battery status prediction.

Description

Battery health state prediction method and device

Technical Field

The disclosure relates to the field of simulation technologies, and in particular, to a method and a device for predicting a battery health state.

Background

In the field of battery management, the prediction and monitoring of the state of health (stateof health, SOH) of a battery is a core task. The accurate prediction of SOH is important for efficient use, production, maintenance and recycling of batteries.

In the related art, when predicting the battery state, the battery can run under a complex working condition, so that the accuracy of predicting the battery state of health is low.

Disclosure of Invention

The disclosure provides a battery state of health prediction method, device and system, which at least solve the problem of low battery state of health accuracy in the related art. The technical scheme of the present disclosure is as follows:

According to a first aspect of embodiments of the present disclosure, there is provided a battery state of health prediction method, including:

acquiring a first training data set, and training a battery state prediction model according to the first training data set to determine a first parameter set, wherein the first training data set is working condition data in battery operation;

Acquiring a second training data set, and training the battery state prediction model according to the second training data set to determine a second parameter set, wherein the second training data set is battery state-of-health related data;

Configuring the first and second parameter sets into the battery state prediction model;

and inputting the real-time parameters of the battery into the battery state prediction model for reasoning operation so as to determine the state quantity of the battery.

Optionally, the battery state prediction model comprises a solid electrolyte interface SEI film generation model and a lithium precipitation reaction model;

In the SEI film generation model, SEI generated current is determined according to SEI film thickness, SEI generated negative electrode particle surface overpotential, SEI film molar volume ratio and a neural network model correction term, wherein the neural network model correction term is obtained after fitting according to main reaction current density, negative electrode potential, temperature and preset coefficients.

Optionally, in the formula of the lithium-precipitation reaction model, the lithium-precipitation current is determined and calculated according to the initial lithium-precipitation current, the temperature, the surface overpotential of the lithium-generated negative electrode particles, the surface overpotential of the lithium-precipitation negative electrode and the neural network model correction term.

Optionally, the working condition data includes current data, voltage data and temperature data corresponding to the battery at a plurality of temperatures, and the training the battery state prediction model to determine the first parameter set includes:

Inputting current data and temperature data in the first training data set into the battery state prediction model for reasoning operation to generate a predicted voltage value;

inputting the predicted voltage value and the corresponding voltage data into a first objective function to calculate a first function value;

and adjusting the numerical value of a first parameter set in the training battery state prediction model with the aim of reducing the first function value, wherein the first parameter set is a non-aging parameter.

Optionally, the inputting the predicted voltage value and the corresponding voltage data into a first objective function to calculate a first function value includes:

and calculating the predicted voltage value corresponding to each battery cell according to the voltage data corresponding to each battery cell so as to obtain the first function value.

Optionally, the second training data set includes actual aging data of the battery during cyclic discharge under multiple working conditions, the obtaining the second training data set, training the battery state prediction model according to the second training data set to determine a second parameter set includes:

inputting the aging data in the second training data set into the battery state prediction model for reasoning operation to generate predicted aging data;

Inputting the predicted aging data and the corresponding actual aging data into a second objective function to calculate a second function value;

And adjusting the numerical value of a second parameter set in the training battery state prediction model with the aim of reducing the second function value, wherein the second parameter set is an aging parameter and a parameter in the neural network model correction term.

8. Optionally, the inputting the predicted aging data and the corresponding actual aging data into a second objective function to calculate a second function value includes:

And inputting a preset weight coefficient, an actual available capacity attenuation value, a predicted direct current internal resistance increase value and an actual direct current internal resistance increase value into the second objective function for calculation to obtain the second function value.

Optionally, the adjusting the value of the second parameter set in the training battery state prediction model with the objective of reducing the second function value includes:

Calculating the change rate of the available capacity attenuation value of each cell along with time;

And when the change rate is larger than a preset change rate threshold value, reducing the step length of extrapolation calculation capacity fading.

Optionally, the aging parameters include:

the SEI film initial thickness, chemical reaction rate, lithium ion diffusion rate in the active material, and initial SEI generation current.

Optionally, the battery state of health quantity includes an available capacity fade, a dc internal resistance increase value, and the aging parameter.

According to a second aspect of embodiments of the present disclosure, there is provided a battery state of health prediction apparatus, including:

the first training module is used for acquiring a first training data set, training a battery state prediction model according to the first training data set to determine a first parameter set, wherein the first training data set is working condition data in battery operation;

The second training module is used for acquiring a second training data set, training the battery state prediction model according to the second training data set to determine a second parameter set, wherein the second training data set is battery health state related data;

A configuration module for configuring the first parameter set and the second parameter set into the battery state prediction model;

And the state prediction module is used for inputting the real-time parameters of the battery into the battery state prediction model to perform reasoning operation so as to determine the state quantity of the battery.

According to a third aspect of the embodiments of the present disclosure, there is provided an electronic device, including:

A processor;

a memory for storing the processor-executable instructions;

Wherein the processor is configured to execute the instructions to implement the method of any of the first aspects.

According to a fourth aspect of embodiments of the present disclosure, there is provided a computer readable storage medium, which when executed by a processor of an electronic device, causes the electronic device to perform the method of any one of the first aspects.

According to a fifth aspect of embodiments of the present disclosure, there is provided a computer program product comprising a computer program which, when executed by a processor, implements the method according to any of the first aspects.

The technical scheme provided by the embodiment of the disclosure at least brings the following beneficial effects:

the parameters in the battery state prediction model are trained through the two training data sets, so that the prediction of the battery state parameters is realized, the problem of insufficient prediction precision is avoided, and the accuracy of the battery state prediction is improved.

The battery state prediction model is trained by utilizing data under various working conditions during training, can cover various cyclic aging working conditions, and the neural network model correction term in the battery state prediction model enriches the model aging simulation mechanism and provides better prediction robustness and accuracy.

The battery state prediction model is a model driven by non-pure data, a large amount of experimental data is not needed for training, the data volume required by training is greatly reduced, the cost is low, and the battery state prediction model is easier to arrange in an embedded system.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the disclosure and together with the description, serve to explain the principles of the disclosure and do not constitute an undue limitation on the disclosure.

Fig. 1 is a flowchart illustrating a battery state of health prediction method according to an exemplary embodiment.

Fig. 2 is a schematic diagram illustrating a structure of a neural network according to an exemplary embodiment.

Fig. 3 is a schematic diagram illustrating a change in a battery state of health parameter according to an exemplary embodiment.

Fig. 4 is a block diagram illustrating a battery state of health prediction apparatus according to an exemplary embodiment.

Detailed Description

In order to enable those skilled in the art to better understand the technical solutions of the present disclosure, the technical solutions of the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings.

It should be noted that the terms "first," "second," and the like in the description and claims of the present disclosure and in the foregoing figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the disclosure described herein may be capable of operation in sequences other than those illustrated or described herein. The implementations described in the following exemplary examples are not representative of all implementations consistent with the present disclosure. Rather, they are merely examples of apparatus and methods consistent with some aspects of the disclosure as detailed in the accompanying claims.

The user information (including but not limited to user equipment information, user personal information, etc.) related to the present disclosure is information authorized by the user or sufficiently authorized by each party.

The prediction and monitoring of SOH is one of the core tasks of conventional Battery management systems (Battery MANAGEMENT SYSTEM, BMS). The accurate prediction of SOH is important for efficient use, production, maintenance and recycling of batteries. For example, the SOH accurate prediction function can be embedded into the production and experiment of the battery in a closed loop manner, so that the product research and development period is shortened, the secondary recycling efficiency of the battery can be improved through monitoring the service life of the battery, and the recycling economy is promoted. Common SOH estimation methods can be broadly divided into experimental analysis methods and modeling simulation methods. Where modeling simulation methods are more widely used due to their lower cost and high efficiency. Engineering technicians often build electrochemical models, equivalent circuit models, empirical aging models, or data-driven AI models to predict SOH. One of the difficulties of the SOH prediction technology is that a common SOH empirical algorithm model cannot meet the prediction accuracy under a complex working condition, and a knee point (knee point) of cliff type aging of a lithium ion battery is difficult to predict.

For the offline prediction model of SOH, it can be generally divided into two technical routes:

In one possible embodiment, a prediction method of a battery model is established to predict SOH of a battery. However, the conventional electrochemical model has the problems of large calculation amount, numerous related parameters and difficult identification, and the model is difficult to cover complex and changeable working conditions. Complicated parameter fitting is needed, an overfitting phenomenon is easy to generate, and the model robustness is insufficient. And it is difficult for a technician to determine the cell aging mechanism without disassembling the battery pack.

In one possible embodiment, the SOH of the battery is predicted using a pure data driven approach, for example, training an LSTM neural network with aging data to predict the SOH of the battery, but this embodiment requires a large amount of data, and the data required for prediction is difficult to obtain and costly.

Fig. 1 is a flowchart illustrating a battery state of health prediction method according to an exemplary embodiment, including the following steps, as shown in fig. 1.

Step 101, a first training data set is obtained, and a battery state prediction model is trained according to the first training data set to determine a first parameter set, wherein the first training data set is working condition data in battery operation;

Step 102, a second training data set is obtained, and the battery state prediction model is trained according to the second training data set to determine a second parameter set, wherein the second training data set is battery state related data;

In this embodiment, the method of combining the electrochemical model with the neural network model is used to predict the health state parameters of the battery. The common electrochemical models include a P2D model, a single particle model (Singlparticlmodel, SPM), an extended single particle model (SPMe) and the like, wherein the P2D model and the SPMe have higher calculation accuracy, can reflect the electrochemical mechanism of the lithium ion battery more accurately, but have larger calculation amount and high calculation cost when being applied to an embedded BMS. Meanwhile, the simulation process of battery aging is a long-time and complex process, so that a single-particle model is used in the embodiment, and the requirement on calculation force can be reduced while higher simulation precision is realized by using the single-particle model.

In the single particle model, in order to describe the aging process of the lithium ion battery, various aging sub-models (comprising an SEI film growth model and a lithium precipitation reaction model) are added together, wherein the aging sub-models are used for calculating aging-related data of the battery, and the maximum concentration of lithium ions in positive and negative electrodes is reduced due to lithium loss caused by SEI film growth and lithium precipitation, so that the battery is aged. In the first charge and discharge process of the liquid lithium ion battery, the electrode material reacts with the electrolyte on the solid-liquid phase interface to form a passivation layer covering the surface of the electrode material. The passivation layer is an interface layer, has the characteristic of solid electrolyte, is an electronic insulator but is an excellent conductor of Li+, and Li+ can be freely inserted and extracted through the passivation layer, so that the passivation film is called a solid electrolyte interface film, SEI film for short.

In the present predictive model, the aging sub-model (including the SEI film growth model and the lithium evolution reaction model) is affected by a relatively single electrochemical mechanism. In fact, aging phenomena such as SEI can be explained by numerous electrochemical principles, each principle has a corresponding mathematical model, and it is difficult to build a pure chemical model to cover numerous aging mechanisms. Therefore, in this embodiment, a hybrid model (battery state prediction model) based on an electrochemical aging mechanism and incorporating a neural network is constructed, and as shown in the figure, a neural network is introduced as a correction term for nonlinearity (denoted by mlp) in calculating the SEI film growth model and the lithium-precipitation reaction model. After a large amount of battery cell aging working condition data are trained, the battery state prediction model can predict SOH of the battery under different working conditions.

Firstly, fitting non-aging parameters of a single particle model through a first training data set, and then training and optimizing the aging parameters and neural network model correction items through a second training data set.

Step 103, configuring the first parameter set and the second parameter set into the battery state prediction model.

And 104, inputting the real-time parameters of the battery into the battery state prediction model for reasoning operation so as to determine the state quantity of the battery.

After the offline training of the battery state prediction model is completed, the model may be deployed into application layer software of the BMS embedded system. Data collected by sensors such as current, voltage, temperature, etc. will be stored as historical information for use by the user. After data cleaning and noise reduction, the historical data can be used as the input of an SOH fusion model to estimate the SOH value of each current battery cell in real time. Meanwhile, according to the daily driving habit of a user, the SOH of the battery cell in the battery in a period of time in the future can be reasonably predicted.

In one possible embodiment, the single particle model abstracts the anode and cathode of the lithium ion battery into two spheres, and simultaneously simplifies the electrolyte, ignoring the concentration and diffusion rate of lithium ions in the electrolyte.

The basic control equation of the single particle model is the feik's law of diffusion, where c _i (r, t) is the concentration of lithium ions in the positive and negative particles, and D _i is the diffusion constant of lithium ions in the particles. The boundary conditions of the single particle model are as follows, when the gradient of lithium ions is calculated on the particle surface for symmetry reasons, i.e. r=r _i, the concentration gradient is then related to the total current density i _i.

The main reaction current density i_i of the positive and negative electrode lithium intercalation and deintercalation can be calculated by the Butler Volmer equation

In addition, the temperature is also a factor with great influence on the aging of the lithium battery, and the model uses a lumped thermal model which comprises three heat sources of ohmic internal resistance heat generation, electrochemical heat generation and convective heat transfer.

For the single particle model we can sum up 36 parameters that need to be identified, as shown in table 1.

TABLE 1

In the SEI film generation model, the surface overpotential of the negative electrode particles generated by SEI according to the SEI film thickness, the molar volume ratio of the SEI film and the neural network model correction term determine SEI generation current.

Alternatively, the formulation of the SEI film generation model is expressed as:

Wherein i _sei is SEI generated current, delta is SEI film thickness, eta _sei is SEI generated negative electrode particle surface overpotential, M is molar volume ratio of SEI film, mlp (j, u, T) is neural network model correction term, j is main reaction current density, u is negative electrode potential, T is temperature, k is preset coefficient, and the neural network model correction term is obtained after fitting according to the main reaction current density, the negative electrode potential, the temperature and the preset coefficient. The SEI film generation model is a model controlled by the chemical reaction rate k _sei and the diffusion rate D _sei of lithium ions in active substances, and the SEI film thickness increment can be obtained by integrating the SEI generation current i _sei. An increase in film thickness implies, on the one hand, a decrease in available capacity due to loss of active lithium ions and, on the other hand, an increase in internal resistance of the battery.

Optionally, in the lithium-precipitation reaction model, the lithium-precipitation current is determined and calculated according to the initial lithium-precipitation current, the temperature, the surface overpotential of the negative electrode particles generated by lithium, the surface overpotential of the negative electrode for lithium precipitation and the neural network model correction term.

Alternatively, the formulation of the lithium-ion reaction model is expressed as:

Where i _pl is the lithium-evolving current, η _pl is the negative electrode surface overpotential for lithium evolution. Similar to the influence of SEI film generation on capacity and internal resistance, the integration of the lithium-precipitation generation current i _pl can also obtain the loss of active lithium caused by irreversible lithium-precipitation, so that the attenuated capacity is calculated according to an electrochemical model.

After the complete electrochemical model is obtained, we can get the simulation values of the current available capacity and the internal DC resistance. The capacity calculation can be obtained from the following formula:

k=n, p represents the negative or positive electrode, respectively, C _k represents the current capacity of the electrode, For the maximum concentration of lithium ions in the positive and negative electrodes, both the increase of the SEI film and the lithium loss caused by lithium precipitation result in the decrease of the maximum concentration. Epsilon _s,k represents the volume fraction of the positive and negative electrode active materials,The actual SOC at full charge and full discharge, respectively, A and L _k are the physical dimensions of the single-particle model.

The calculation of the direct current internal resistance can be obtained by a mixed pulse power performance test (Hybrid Pulse Power Characterization, HPPC) simulation test, and the capacity simulation array and the internal resistance simulation value are respectively marked as Cap _sim and R _sim.

In this embodiment, a neural network model correction term mlp (j, u, T) is added to both the SEI film generation model and the lithium-ion reaction model.

Fig. 2 is a schematic diagram of a neural network according to an exemplary embodiment, as shown in fig. 2, and in one possible embodiment, the neural network is of the type Multi-layer perceptron (Multi-layer perceptron). The neural network comprises three layers, namely an input layer, a hidden layer and an output layer. The input layer, denoted as X, contains three elements, denoted as X (j, u, T), and the input variables are the main reaction current density j, the negative potential u, and the temperature T, respectively. The hidden layer H contains 64 elements, a nonlinear activation function sigmoid is added in the hidden layer to carry out nonlinear mapping, and then the nonlinear mapping is carried out on the output layer O. The calculated relationship is as follows. The output value O, i.e. k mlp (j, u, T) can be regarded as a correction value for i _sei or i _pl.

The functions in the neural network are as follows:

H=W₁X+b₁;

H′=sigmoid(H);

O=W₂H'+b₂。

In one possible embodiment, the first training data set comprises 3 lithium iron phosphate battery cells at 0 ℃,10 ℃,25 ℃,40 ℃ CLTC, WLTC, and US06 operating condition data. The total of 36 groups of data are combined in a crossing way, and each group of data comprises dynamic working condition current, voltage and temperature data. The real voltage is denoted as U _real, and the data set is divided into a training data set 1 and a test data set 1 according to the proportion of 8:2, wherein the training data set is used for training a model, and the test data set is used for testing the accuracy of the prediction of the trained model.

Optionally, the step of inputting the predicted voltage value and the corresponding voltage data into a first objective function to calculate a first function value includes calculating the predicted voltage value corresponding to each cell according to the voltage data corresponding to each cell to obtain the first function value.

Alternatively, the formulation of the first objective function is expressed as:

Wherein Fitnessfunction is the first objective function, U _real is the voltage data, U _sim is the predicted voltage value, rmse (d) is a root mean square error function, i is the cell number in the current battery, and n is the maximum cell number in the battery.

In one possible embodiment, the second training data set comprises test data for 12 lithium iron phosphate cells at 0 ℃,10 ℃,25 ℃,40 ℃, with the cyclic discharge process being CLTC and US06 conditions, the charging process being a 1C constant current charge, and the intersecting combinations being 96 sets of data. After every 40 cycles, SOH detection is carried out, wherein the detection comprises a 1/3C constant volume test and an HPPC internal resistance test, an actual capacity attenuation vector Cap _real and a current actual internal resistance value R _real are respectively obtained, and each group of experiments is finished when the available capacity is attenuated to 80%. Data set 2 is also divided into training data set 2, which is used to train the model, and test data set 2, which is used to test the accuracy of the trained model predictions, in an 8:2 ratio.

Optionally, the method for training the first parameter set and the second parameter set in the battery state prediction model is a particle swarm optimization (particleswarmoptimization, PSO) algorithm, which is a spatial random search algorithm for solving the optimal solution problem. The 20 particles set during each iteration are moved towards the current objective function optimum until they are moved to the boundary or the maximum number of iterations is reached. The particle flight speed and the current position can be calculated by the following formula:

Optionally, the aging data includes an available capacity attenuation value and a dc internal resistance increase value, and said inputting the predicted aging data and the corresponding actual aging data into a second objective function to calculate a second function value includes:

Optionally, the second objective function is formulated as:

Wherein Fitness function2 is the second objective function, abs () is an absolute value calculation function, w is a weight coefficient, cap _real is an actual available capacity attenuation value, cap _sim is a predicted available capacity attenuation value, deltaR _sim is a predicted DC internal resistance increase value, deltaR _real is an actual DC internal resistance increase value.

Alternatively, the value of w is set to 0.5.

And when the change rate is larger than a preset change rate threshold, reducing the step length of extrapolation calculation capacity attenuation.

FIG. 3 is a schematic diagram illustrating a change in a battery state of health parameter, as shown in FIG. 3, using an adaptive step-size extrapolation algorithm for the aging simulation process in training the second parameter set in the model to ensure accuracy and high efficiency of the aging simulation, according to an exemplary embodiment.

SOH related tests often run thousands of cycles, with the cost of performing simulation operations one by one being prohibitive. The model adopts an extrapolation algorithm of self-adaptive step length in simulation, namely, when the current cycle capacity attenuation is judged to be too fast, the step length of extrapolation calculation capacity attenuation is reduced in a self-adaptive mode, extrapolation variables are SEI film thickness, irreversible lithium precipitation loss amount and the like, so that finer SOH prediction in a short period can be conveniently performed, and when the capacity attenuation slope is not changed greatly, the simulation step length is increased appropriately, and therefore SOH simulation speed is increased.

Optionally, the aging parameters include an SEI film initial thickness delta ₀, a chemical reaction rate k _sei, a lithium ion diffusion rate D _sei in the active material, and an initial SEI generation current i _0,pl.

Fig. 4 is a block diagram of a battery state of health prediction apparatus according to an exemplary embodiment. Referring to fig. 4, the apparatus includes:

a first training module 410, configured to obtain a first training data set, and train a battery state prediction model according to the first training data set to determine a first parameter set, where the first training data set is working condition data in battery operation;

A second training module 420, configured to obtain a second training data set, and train the battery state prediction model according to the second training data set to determine a second parameter set, where the second training data set is battery health state related data;

A configuration module 430 for configuring the first parameter set and the second parameter set into the battery state prediction model;

the state prediction module 440 is configured to input the real-time parameters of the battery into the battery state prediction model for performing an inference operation to determine the state of health of the battery.

The specific manner in which the various modules perform the operations in the apparatus of the above embodiments have been described in detail in connection with the embodiments of the method, and will not be described in detail herein.

In order to realize the embodiment, the application also provides electronic equipment which comprises a processor and a memory which is in communication connection with the processor, wherein the memory stores computer execution instructions, and the processor executes the computer execution instructions stored in the memory so as to realize the method for executing the embodiment.

In order to implement the above-described embodiments, the present application also proposes a computer-readable storage medium having stored therein computer-executable instructions that, when executed by a processor, are adapted to implement the methods provided by the foregoing embodiments.

In order to implement the above embodiments, the present application also proposes a computer program product comprising a computer program which, when executed by a processor, implements the method provided by the above embodiments.

The processing of collecting, storing, using, processing, transmitting, providing, disclosing and the like of the personal information of the user in the application accords with the regulations of related laws and regulations and does not violate the popular regulations of the public order.

It should be noted that personal information from users should be collected for legitimate and reasonable uses and not shared or sold outside of these legitimate uses. In addition, such collection/sharing should be performed after receiving user informed consent, including but not limited to informing the user to read user agreements/user notifications and signing agreements/authorizations including authorization-related user information before the user uses the functionality. In addition, any necessary steps are taken to safeguard and ensure access to such personal information data and to ensure that other persons having access to the personal information data adhere to their privacy policies and procedures.

The present application contemplates embodiments that may provide a user with selective prevention of use or access to personal information data. That is, the present disclosure contemplates that hardware and/or software may be provided to prevent or block access to such personal information data. Once personal information data is no longer needed, risk can be minimized by limiting data collection and deleting data. In addition, personal identification is removed from such personal information, as applicable, to protect the privacy of the user.

In the foregoing description of embodiments, reference has been made to the terms "one embodiment," "some embodiments," "example," "a particular example," or "some examples," etc., meaning that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the application. In this specification, schematic representations of the above terms are not necessarily directed to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, the different embodiments or examples described in this specification and the features of the different embodiments or examples may be combined and combined by those skilled in the art without contradiction.

Furthermore, the terms "first," "second," and the like, are used for descriptive purposes only and are not to be construed as indicating or implying a relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include at least one such feature. In the description of the present application, the meaning of "plurality" means at least two, for example, two, three, etc., unless specifically defined otherwise.

Any process or method descriptions in flow charts or otherwise described herein may be understood as representing modules, segments, or portions of code which include one or more executable instructions for implementing specific logical functions or steps of the process, and additional implementations are included within the scope of the preferred embodiment of the present application in which functions may be executed out of order from that shown or discussed, including substantially concurrently or in reverse order from that shown or discussed, depending on the functionality involved, as would be understood by those reasonably skilled in the art of the embodiments of the present application.

Logic and/or steps represented in the flowcharts or otherwise described herein, e.g., a ordered listing of executable instructions for implementing logical functions, can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For the purposes of this description, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer-readable medium would include an electrical connection (an electronic device) having one or more wires, a portable computer diskette (a magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer readable medium may even be paper or other suitable medium on which the program is printed, as the program may be electronically captured, via, for instance, optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner, if necessary, and then stored in a computer memory.

It is to be understood that portions of the present application may be implemented in hardware, software, firmware, or a combination thereof. In the above-described embodiments, the various steps or methods may be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. If implemented in hardware as in another embodiment, may be implemented using any one or combination of techniques known in the art, discrete logic circuits with logic gates for implementing logic functions on data signals, application specific integrated circuits with appropriate combinational logic gates, programmable Gate Arrays (PGAs), field Programmable Gate Arrays (FPGAs), etc.

Those of ordinary skill in the art will appreciate that all or a portion of the steps carried out in the method of the above-described embodiments may be implemented by a program to instruct related hardware, where the program may be stored in a computer readable storage medium, and where the program, when executed, includes one or a combination of the steps of the method embodiments.

In addition, each functional unit in the embodiments of the present application may be integrated in one processing module, or each unit may exist alone physically, or two or more units may be integrated in one module. The integrated modules may be implemented in hardware or in software functional modules. The integrated modules may also be stored in a computer readable storage medium if implemented in the form of software functional modules and sold or used as a stand-alone product.

The above-mentioned storage medium may be a read-only memory, a magnetic disk or an optical disk, or the like. While embodiments of the present application have been shown and described above, it will be understood that the above embodiments are illustrative and not to be construed as limiting the application, and that variations, modifications, alternatives and variations may be made to the above embodiments by one of ordinary skill in the art within the scope of the application.

Claims

1. A method for predicting a battery health state, comprising:

Acquire a first training data set, and train a battery state prediction model according to the first training data set to determine a first parameter set, wherein the first training data set is operating condition data of the battery during operation;

Acquire a second training data set, and train the battery state prediction model according to the second training data set to determine a second parameter set, wherein the second training data set is battery health state related data;

configuring the first parameter set and the second parameter set into the battery state prediction model;

The real-time parameters of the battery are input into the battery state prediction model for inference operation to determine the battery health state quantity.

2. The method according to claim 1, characterized in that the battery state prediction model includes a solid electrolyte interface SEI film formation model and a lithium precipitation reaction model;

Among them, in the SEI film generation model, the SEI generation current is determined by the neural network model correction term based on the SEI film thickness, the surface overpotential of the negative electrode particles generated by SEI, the molar volume ratio of the SEI film, and the neural network model correction term, wherein the neural network model correction term is obtained by fitting according to the main reaction current density, the negative electrode potential, the temperature, and the preset coefficient.

3. The method according to claim 2 is characterized in that, in the formula of the lithium deposition reaction model, the lithium deposition current is determined and calculated based on the initial lithium deposition current, temperature, the surface overpotential of the negative electrode particles generated by lithium, the surface overpotential of the negative electrode for lithium deposition and the correction term of the neural network model.

4. The method according to claim 2, characterized in that the operating condition data includes current data, voltage data and temperature data corresponding to the battery at multiple temperatures, and the training of the battery state prediction model to determine the first parameter set includes:

Inputting the current data and the temperature data in the first training data set into the battery state prediction model for inference operation to generate a predicted voltage value;

The values of the first parameter set in the training battery state prediction model are adjusted with the goal of reducing the first function value, wherein the first parameter set is non-aging parameters.

5. The method according to claim 4, characterized in that the step of inputting the predicted voltage value and the corresponding voltage data into a first objective function to calculate a first function value comprises:

According to the voltage data corresponding to each battery cell, the predicted voltage value corresponding to each battery cell is calculated to obtain the first function value.

6. The method according to claim 5, characterized in that the second training data set includes actual aging data of the battery during cyclic discharge under various working conditions, and the obtaining of the second training data set and training the battery state prediction model according to the second training data set to determine the second parameter set comprises:

Inputting the aging data in the second training data set into the battery state prediction model for inference operation to generate predicted aging data;

The values of the second parameter set in the trained battery state prediction model are adjusted with the goal of reducing the second function value, wherein the second parameter set includes aging parameters and parameters in the neural network model correction term.

7. The method according to claim 6, characterized in that the step of inputting the predicted aging data and the corresponding actual aging data into a second objective function to calculate a second function value comprises:

The preset weight coefficient, the actual available capacity attenuation value, the predicted available capacity attenuation value, the predicted DC internal resistance growth value and the actual DC internal resistance growth value are input into the second objective function for calculation to obtain the second function value.

8. The method according to claim 6, characterized in that the step of adjusting the value of the second parameter set in the training battery state prediction model with the goal of reducing the second function value comprises:

Calculate the rate of change of the available capacity attenuation value of each battery cell over time;

When the rate of change is greater than a preset rate of change threshold, the step size of the extrapolated calculation of capacity decay is reduced.

9. The method according to claim 6, wherein the aging parameters include:

The initial thickness of the SEI film, the chemical reaction rate, the diffusion rate of lithium ions in the active material and the initial SEI generation current.

10. The method according to any one of claims 1-9, characterized in that the battery health status includes available capacity attenuation, DC internal resistance growth value and the aging parameter.

11. A battery health status prediction device, comprising:

A first training module is used to obtain a first training data set, and train a battery state prediction model according to the first training data set to determine a first parameter set, wherein the first training data set is operating condition data of the battery during operation;

A second training module is used to obtain a second training data set, and train the battery state prediction model according to the second training data set to determine a second parameter set, wherein the second training data set is battery health state related data;

A configuration module, configured to configure the first parameter set and the second parameter set into the battery state prediction model;

The state prediction module is used to input the real-time parameters of the battery into the battery state prediction model for inference calculation to determine the battery health state.

12. An electronic device, comprising:

processor;

a memory for storing instructions executable by the processor;

The processor is configured to execute the instructions to implement the method according to any one of claims 1 to 10.

13. A computer-readable storage medium, when instructions in the storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the method according to any one of claims 1 to 10.

14. A computer program product comprising a computer program, which, when executed by a processor, implements the method according to any one of claims 1 to 10.