Battery health state prediction method and system

Abstract

The invention provides a battery health state prediction method and system, and the method comprises the steps: obtaining a high-dimensional data set in a battery discharge process, and carrying out the normalization processing of the high-dimensional data set, and obtaining standardized data; analyzing the correlation between the standardized data and the battery capacity by using a Spearman correlation coefficient to obtain a target data set; inputting the target data set into a width auto-encoder, extracting feature nodes and enhanced nodes of the width auto-encoder, and obtaining intermediate feature representation; and inputting the intermediate feature representation into a random configuration network fused with an intuitionistic fuzzy mechanism, and generating a battery health state preliminary prediction result, thereby alleviating the technical problem of low accuracy of the existing battery health state prediction.

Description

Technical Field

[0001] This application relates to the technical field of battery state of health prediction, and particularly to a method and system for predicting battery state of health. Background Art

[0002] With the wide application of electric vehicles (EVs), renewable energy storage systems, and portable electronic devices, the demand for batteries as key components is increasing day by day. The state of health (SOH) of a battery is one of the important indicators for measuring battery performance and life.

[0003] Accurately predicting the SOH of a battery is of great significance for extending battery life, improving system reliability, and optimizing energy management. However, due to the complex electrochemical reaction process inside the battery and its large influence by environmental factors, traditional methods based on empirical formulas or simple statistical models are difficult to provide high-precision SOH prediction results. Summary of the Invention

[0004] This application provides a method and system for predicting battery state of health to solve the problem of low accuracy in existing battery state of health prediction.

[0005] In a first aspect, this application provides a method for predicting battery state of health, including: obtaining a high-dimensional data set during the battery discharge process, and performing normalization processing on the high-dimensional data set to obtain standardized data, where the high-dimensional data set at least includes: current data, voltage data, temperature data, battery capacity data, battery internal resistance data, battery self-discharge rate; analyzing the correlation between the standardized data and the battery capacity using the Spearman correlation coefficient to obtain a target data set, where the data included in the target data set are the standardized data with a Spearman correlation coefficient greater than a preset threshold; inputting the target data set into a width autoencoder, extracting the feature nodes and enhancement nodes of the width autoencoder to obtain an intermediate feature representation; inputting the intermediate feature representation into a random configuration network integrated with an intuitionistic fuzzy mechanism to generate a preliminary prediction result of the battery state of health.

[0006] Further, input the target data set into a width autoencoder, extract the feature nodes and enhancement nodes of the width autoencoder, and obtain an intermediate feature representation, including: using the mapping layer and the random weight matrix of the width autoencoder to convert the target data set into feature nodes to obtain a first-stage feature representation; expanding the first-stage feature representation into the enhancement nodes of the width autoencoder to form a second-stage feature representation; using the reconstruction layer of the width autoencoder, the second-stage feature representation, and the linear weights to reconstruct the target data set to obtain reconstructed data; minimizing the mean square error between the reconstructed data and the target data set to obtain the intermediate feature representation.

[0007] Further, input the intermediate feature representation into a random configuration network integrated with an intuitionistic fuzzy mechanism to generate a preliminary prediction result of the battery health state, including: initializing the random configuration network structure; inputting the intermediate feature representation into the initialized random configuration network structure to obtain an initial network output; using the initial network output and the intuitionistic fuzzy mechanism to iteratively optimize the initialized random configuration network structure to obtain an optimized random configuration network structure; calculating the preliminary prediction result of the battery health state based on the optimized random configuration network structure.

[0008] Further, the method further includes: combining the physical constraint conditions in the physical information neural network framework into the optimized random configuration network structure to obtain a target model, and setting a loss function for the target model, where the loss function includes: a data loss function, a physical equation loss function, and a monotonicity loss function; calculating a loss function value based on the loss function and the preliminary prediction result of the battery health state; using the loss function value to iteratively optimize the target model to obtain an optimized target model; using the optimized target model to determine the final prediction result of the battery health state.

[0009] Further, obtaining a high-dimensional data set during the battery discharge process includes: recording the operation data of the battery at different cycle numbers according to a preset battery charge and discharge experiment plan, where the preset battery charge and discharge experiment plan includes a charging strategy and a discharging strategy; performing preliminary cleaning and preprocessing on the operation data to obtain the high-dimensional data set.

[0010] Further, analyzing the correlation between the standardized data and the battery capacity using the Spearman correlation coefficient to obtain a target data set, including: analyzing the correlation between the standardized data and the battery capacity using the Spearman correlation coefficient, and calculating the Spearman correlation coefficient corresponding to each standardized data; constructing the target data set using the standardized data with the Spearman correlation coefficient greater than a preset threshold.

[0011] Further, after obtaining the final prediction result, the method further includes: obtaining the actual result of the battery health state; calculating an error index by using the actual result and the final prediction result, where the error index includes: MAPE value and RMSE value; optimizing the optimized target model by using the error index.

[0012] In a second aspect, the present application provides a battery health state prediction system, including: an acquisition module, configured to acquire a high-dimensional data set during the battery discharge process, and perform normalization processing on the high-dimensional data set to obtain standardized data, where the high-dimensional data set at least includes: current data, voltage data, temperature data, battery capacity data, battery internal resistance data, battery self-discharge rate; an analysis module, configured to analyze the correlation between the standardized data and the battery capacity by using the Spearman correlation coefficient to obtain a target data set, where the data included in the target data set is the standardized data with a Spearman correlation coefficient greater than a preset threshold; an extraction module, configured to input the target data set into a width autoencoder, and extract the feature nodes and enhancement nodes of the width autoencoder to obtain an intermediate feature representation; a generation module, configured to input the intermediate feature representation into a random configuration network integrated with an intuitionistic fuzzy mechanism to generate a preliminary prediction result of the battery health state.

[0013] In a third aspect, the present application provides a computing device, including a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement the method described in the first aspect above.

[0014] In a fourth aspect, the present application provides a computer storage medium, storing a computer program, where when the computer program is executed by a computer, it implements the method described in the first aspect above.

[0015] In the present invention, by acquiring a high-dimensional data set during the battery discharge process, performing normalization processing on the high-dimensional data set to obtain standardized data; analyzing the correlation between the standardized data and the battery capacity by using the Spearman correlation coefficient to obtain a target data set; inputting the target data set into a width autoencoder, and extracting the feature nodes and enhancement nodes of the width autoencoder to obtain an intermediate feature representation; inputting the intermediate feature representation into a random configuration network integrated with an intuitionistic fuzzy mechanism to generate a preliminary prediction result of the battery health state, the purpose of accurately predicting the battery health state is achieved, and further the technical problem of low accuracy of the existing battery health state prediction is solved.

[0016] These aspects or other aspects of the present application will be more clearly understood in the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] To more clearly illustrate the technical solutions in the present application or the prior art, the following will briefly introduce the drawings required for the description of the prior art. Obviously, the drawings in the following description are some of those of the present application. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings without creative efforts.

[0018] Figure 1 It is a flowchart of a method for predicting the state of health of a battery provided by the present application; Figure 2 It is a schematic diagram of a system for predicting the state of health of a battery provided by the present application; Figure 3 It is a schematic diagram of a computing device provided by the present application. Detailed implementation manners

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions of the present invention with reference to the accompanying drawings. Obviously, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0020] Embodiment 1. According to an embodiment of the present invention, an embodiment of a method for predicting the state of health of a battery is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. And although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0021] Figure 1 It is a flowchart of a method for predicting the state of health of a battery according to an embodiment of the present invention. As Figure 1 shown, the method includes the following steps: Step S101, obtain a high-dimensional data set during the battery discharge process, and perform normalization processing on the high-dimensional data set to obtain standardized data, where the high-dimensional data set at least includes: current data, voltage data, temperature data, battery capacity data, battery internal resistance data, battery self-discharge rate; In the embodiment of the present invention, before starting to predict the state of health of the battery, it is first necessary to obtain a high-dimensional data set during the battery discharge process. These data include but are not limited to current data, voltage data, temperature data, battery capacity data, battery internal resistance data, and battery self-discharge rate. These data can be collected through a battery management system (BMS) or other sensor devices.

[0022] Specifically, obtaining a high-dimensional data set during the battery discharge process includes the following steps: According to a preset battery charge-discharge experiment plan, record the operation data of the battery at different cycle numbers, where the preset battery charge-discharge experiment plan includes a charging strategy and a discharging strategy; Perform preliminary cleaning and preprocessing on the operation data to obtain the high-dimensional data set.

[0023] Before obtaining the high-dimensional data set during the battery discharge process, it is first necessary to record the operation data of the battery at different cycle numbers according to a preset battery charge-discharge experiment plan. This experiment plan should include a charging strategy and a discharging strategy to ensure the comprehensiveness and representativeness of the data.

[0024] Charging strategy: Constant current charging (CC), constant voltage charging (CV), or a mixed charging strategy can be selected. For example, during the constant current charging stage, the current remains constant until the voltage reaches the set value; during the constant voltage charging stage, the voltage remains constant and the current gradually decreases until it stops.

[0025] Discharging strategy: Constant current discharging (CC) or a pulse discharging strategy can be selected. For example, during the constant current discharging process, the current remains constant until the battery voltage drops to the minimum allowable value.

[0026] During the experiment, use a battery management system (BMS) or other sensor devices to record various parameters of the battery in real time, including but not limited to current, voltage, temperature, battery capacity, battery internal resistance, and self-discharge rate, etc. These data will serve as the basis for subsequent analysis.

[0027] To ensure the quality and consistency of the data, it is necessary to perform preliminary cleaning and preprocessing on the collected data. Common processing methods include removing missing values, smoothing, and filtering, etc.

[0028] Removing missing values: For data points with missing values, interpolation or deletion methods can be used for processing. The interpolation method fills in the missing values through linear interpolation of adjacent data points; the deletion method directly deletes the data points containing missing values.

[0029] Smoothing: To reduce the influence of noise, a moving average filter or a low-pass filter can be used to smooth the data.

[0030] Filtering: A median filter can be used to remove outliers. The median filter sorts the data points within the window and takes the middle value as the output value, thereby effectively removing extreme values.

[0031] To eliminate the influence of different dimensions on model training, it is necessary to normalize the processed data. Commonly used normalization methods include min-max normalization and Z-score standardization. The present invention adopts the min-max normalization method to map the data to the interval [0, 1].

[0032] Step S102: Analyze the correlation between the standardized data and the battery capacity using the Spearman correlation coefficient to obtain a target data set, where the data included in the target data set are the standardized data with a Spearman correlation coefficient greater than a preset threshold; Specifically, step S102 includes the following steps: Analyze the correlation between the standardized data and the battery capacity using the Spearman correlation coefficient, and calculate the Spearman correlation coefficient corresponding to each standardized data; Use the standardized data with a Spearman correlation coefficient greater than the preset threshold to construct the target data set.

[0033] To improve the accuracy and efficiency of the model, it is necessary to select features highly correlated with the battery capacity from the high-dimensional data set. The present invention uses the Spearman rank correlation coefficient to evaluate the correlation between each feature and the battery capacity.

[0034] For each standardized data, calculate the Spearman correlation coefficient between it and the battery capacity respectively.

[0035] According to the calculated Spearman correlation coefficient, screen out the standardized data greater than the preset threshold to construct a target data set. This set will serve as the basis for subsequent feature extraction and model training.

[0036] Specifically, the core index of the state of health (SOH) of the battery is the battery capacity, and there are significant differences in the correlation between multi-dimensional parameters (such as current, voltage, temperature, etc.) during the charge and discharge process and capacity attenuation. The Spearman correlation coefficient can effectively measure non-linear statistical relationships. By threshold screening, irrelevant or weakly correlated parameters can be eliminated, and core features highly correlated with the capacity (such as voltage slope, current dynamic characteristics, etc.) can be retained, so as to focus on the key physical factors affecting battery degradation and avoid interference from redundant information in model learning.

[0037] Furthermore, battery aging is essentially a process of non-linear degradation of electrochemical parameters (such as active material loss, lithium deposition, etc.). The Spearman correlation coefficient can capture the non-linear relationship between parameters and capacity, which is superior to the Pearson coefficient that only measures linear relationships. For example, the non-linear change of the voltage curve is closely related to capacity attenuation, but it may not be fully characterized by linear correlation. Threshold screening ensures that the input data conforms to the physical laws of battery degradation and provides an appropriate feature representation for the subsequent neural network integrating physical constraints.

[0038] In addition, high-dimensional battery data often contains noise or low-information parameters (such as instantaneous fluctuations under certain working conditions). Through threshold filtering, the data dimension can be significantly reduced, the sensitivity of the model to noise can be decreased, and the generalization ability can be enhanced. Meanwhile, after feature dimensionality reduction, dynamic structures such as the bottleneck autoencoder (BAE) and the sparse coding network (SCN) can extract hierarchical features more efficiently, avoiding overfitting or waste of computing resources caused by redundant inputs, and achieving high-precision prediction under limited data conditions.

[0039] In summary, by screening highly correlated features, the prediction error can be reduced, and at the same time, the model training time can be decreased, achieving efficient and stable prediction of battery health state while ensuring accuracy.

[0040] Step S103: Input the target data set into the bottleneck autoencoder, extract the feature nodes and enhancement nodes of the bottleneck autoencoder, and obtain an intermediate feature representation. Specifically, step S103 includes the following steps: Use the mapping layer and the random weight matrix of the bottleneck autoencoder to transform the target data set into feature nodes, and obtain a first-stage feature representation. Expand the first-stage feature representation into the enhancement nodes of the bottleneck autoencoder to form a second-stage feature representation. Use the reconstruction layer of the bottleneck autoencoder, the second-stage feature representation, and the linear weights to reconstruct the target data set to obtain reconstructed data. Minimize the mean square error between the reconstructed data and the target data set to obtain the intermediate feature representation.

[0041] The bottleneck autoencoder (BAE) is a deep learning model used to extract useful feature representations from input data. First, initialize the structure of the bottleneck autoencoder, including defining the mapping layer, the expansion layer, and the reconstruction layer. The mapping layer is used to transform the input data into feature nodes, the expansion layer is used to increase the network width to form enhancement nodes, and the reconstruction layer is used to reconstruct the input data.

[0042] Use the mapping layer and the random weight matrix to transform the target data set into feature nodes: Through the random weight matrix and the bias term perform a linear transformation, and use an activation function (such as ReLU) for non-linear transformation to obtain a first-stage feature representation .

[0043] Through the weight matrix of the enhancement layer and the bias term perform a linear transformation, and use an activation function (such as ReLU) for non-linear transformation to obtain enhancement nodes .

[0044] Concatenate the splicing feature nodes and enhancement nodes [Z∣H], and through the weight matrix of the reconstruction layer perform a linear transformation to obtain the reconstructed data .

[0045] Finally, minimize the mean square error between the reconstructed data and the target data set, and optimize the network parameters , and obtain the final feature representation as the intermediate feature representation.

[0046] In order to effectively extract useful features from the original data, this application designs a width autoencoder including a mapping layer, an expansion layer, and a reconstruction layer. The target data set is converted into feature nodes through the mapping layer and a random weight matrix, and then expanded into enhancement nodes, increasing the network width and enhancing the expression ability. The reconstruction layer is used to reconstruct the feature representation in the second stage, and the reconstruction error is minimized to ensure the effectiveness and stability of feature extraction, not only simplifying the feature selection process, but also improving the performance and reliability of the model.

[0047] Step S104: Input the intermediate feature representation into a random configuration network integrated with an intuitionistic fuzzy mechanism to generate a preliminary prediction result of the battery health state.

[0048] In the embodiment of the present invention, step S104 includes the following steps: Initialize the structure of the random configuration network; Input the intermediate feature representation into the initialized random configuration network structure to obtain an initial network output; Use the initial network output and the intuitionistic fuzzy mechanism to iteratively optimize the initialized random configuration network structure to obtain an optimized random configuration network structure; Based on the optimized random configuration network structure, calculate the preliminary prediction result of the battery health state.

[0049] Specifically, the random configuration network (SCN) is a neural network model for dealing with complex non-linear problems. First, initialize the structure of the SCN, including setting the number of initial hidden layer nodes. Then, take the intermediate feature representation as the input and calculate the initial network output through forward propagation.

[0050] In each iteration process, use the current network output to compare with the objective function, and calculate the residual vector , where is the objective function value, is the current network output. By calculating the residual vector, evaluate the difference between the current network output and the objective function to obtain the residual analysis result.

[0051] Use the intuitionistic fuzzy strategy to evaluate the contribution degree of each candidate node. Nodes with lower contribution degrees will be automatically filtered out, thus simplifying the network structure and improving efficiency. This process helps to dynamically adjust the network structure to make it more suitable for the current task.

[0052] Recalculate the loss function according to the updated network structure and dynamically adjust the loss weights. This step ensures that the model can more accurately predict the state of health (SOH) of the battery. Based on the adjusted loss weights, calculate the new current network output through forward propagation and iterate the above process until the preset stop condition is met to obtain the preliminary prediction result of the battery state of health.

[0053] To further accurately calculate the predicted value of the battery state of health (SOH), combine the physical constraint conditions under the physics-informed neural network (PINN) framework, and set the total loss function including data loss, physical equation loss, and monotonicity loss. Use the cosine annealing strategy to adjust the learning rate to help the model jump out of the local optimal solution, and iterate to train the model until the preset stop condition is reached. Finally, use the optimized model to determine the final prediction result of the battery state of health. This method not only improves the generalization ability and interpretability of the model but also maintains high stability and reliability under different working conditions.

[0054] Furthermore, after obtaining the preliminary prediction result of the battery state of health, the embodiment of the present invention further includes the following steps: Combine the physical constraint conditions under the physics-informed neural network framework into the optimized random configuration network structure to obtain a target model, and set a loss function for the target model, where the loss function includes: a data loss function, a physical equation loss function, and a monotonicity loss function; Based on the loss function and the preliminary prediction result of the battery state of health, calculate the loss function value; Use the loss function value to iteratively optimize the target model to obtain an optimized target model; Use the optimized target model to determine the final prediction result of the battery state of health.

[0055] To further accurately calculate the predicted value of the battery state of health (SOH), combine the physical constraint conditions under the physics-informed neural network (PINN) framework. First, set the total loss function, which includes data loss, physical equation loss, and monotonicity loss.

[0056] To help the model jump out of the local optimum, a cosine annealing strategy is adopted to adjust the learning rate. Then, model training is carried out until the preset stopping condition is reached, and the optimized model parameters are obtained for calculating the final SOH prediction value. The test set data is verified, and the final SOH prediction value is calculated using the optimized model parameters, and the model performance metrics such as MAE, MSE, etc. are evaluated.

[0057] After obtaining the said final prediction result, the embodiments of the present invention further include the following steps: Obtain the actual result of the battery health state; Using the said actual result and the said final prediction result, calculate the error metrics, wherein, the error metrics include: MAPE value and RMSE value; Optimize the said optimized target model using the said error metrics.

[0058] After obtaining the final prediction result, it is necessary to obtain the actual result of the battery health state. The actual result can be obtained through experimental measurement or historical data. For example, the actual capacity change of the battery can be recorded by the battery management system (BMS).

[0059] Using the actual result and the final prediction result, calculate the error metrics. Commonly used error metrics include mean square error (MSE), mean absolute error (MAE), root mean square error (RMSE), and mean absolute percentage error (MAPE) Analyze the error sources using the error metrics to identify the factors that may cause the error to increase. Common error sources include data quality problems: If the errors are mainly concentrated in certain specific time periods or specific conditions, it may be caused by data quality problems. For example, sensor failures or environmental disturbances may introduce outliers.

[0060] Unreasonable model structure: If the errors are large and evenly distributed, it may be caused by an unreasonable model structure. For example, the model is too simple to capture complex relationships, or the model is too complex resulting in overfitting.

[0061] Improper hyperparameter setting: If the errors fluctuate greatly over time, it may be caused by improper hyperparameter setting. For example, too large a learning rate may cause the model to not converge, and too small a learning rate may cause too long a training time. The sources include data quality problems, unreasonable model structure, improper hyperparameter setting, etc.

[0062] After adjusting the model parameters or improving the model structure, the training and verification processes need to be repeated until the predetermined accuracy requirement is met, and an improved model is obtained. Specific verification methods include cross-validation, hold-out method, and bootstrap method, etc.

[0063] To further optimize the model performance, after obtaining the final prediction result, the actual result of the battery health state is acquired, and error metrics (such as MAPE, RMSE, etc.) are calculated using the actual result and the prediction result. Factors that may cause an increase in errors are identified through error analysis, and the model parameters or the model structure are adjusted accordingly. The training and validation processes are repeated until the predetermined accuracy requirements are met. This method ensures that the model can achieve the best performance after multiple optimizations, improving the practical value and reliability of the model.

[0064] The present invention proposes a method for predicting the battery health state based on an intuitionistic fuzzy random collocation physics-informed neural network. This method obtains a high-dimensional data set during the battery discharge process, analyzes the correlation between the standardized data and the battery capacity using the Spearman correlation coefficient, extracts feature nodes and enhancement nodes, and combines a random collocation network (SCN) and a physics-informed neural network (PINN) to predict the battery health state (SOH). This method has high accuracy and robustness and is suitable for practical applications.

[0065] Embodiment 2, The embodiment of the present invention also provides a battery health state prediction system, which is used to execute the battery health state prediction method provided in the above content of the embodiment of the present invention. The following is a specific introduction to the battery health state prediction system provided by the embodiment of the present invention.

[0066] As Figure 2 shown, Figure 2 is a schematic diagram of the above battery health state prediction system, and the system includes: An acquisition module 10, configured to acquire a high-dimensional data set during the battery discharge process and perform normalization processing on the high-dimensional data set to obtain standardized data, where the high-dimensional data set at least includes: current data, voltage data, temperature data, battery capacity data, battery internal resistance data, and battery self-discharge rate; An analysis module 20, configured to analyze the correlation between the standardized data and the battery capacity using the Spearman correlation coefficient to obtain a target data set, where the data included in the target data set are standardized data with a Spearman correlation coefficient greater than a preset threshold; An extraction module 30, configured to input the target data set into a width autoencoder, extract the feature nodes and enhancement nodes of the width autoencoder, and obtain an intermediate feature representation; A generation module 40, configured to input the intermediate feature representation into a random collocation network integrated with an intuitionistic fuzzy mechanism to generate a preliminary prediction result of the battery health state.

[0067] Embodiment 3, The embodiment of the present invention also provides a computing device for executing the program of the method described in Embodiment 1 above, such as Figure 3As shown, the computing device may include a storage component 41 and a processing component 42; The storage component 41 stores one or more computer instructions, and the one or more computer instructions are called and executed by the processing component 42.

[0068] The processing component 42 may include one or more processors to execute computer instructions to complete all or part of the steps in the method of Embodiment 1. Of course, the processing component may also be implemented by one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components for executing the above method.

[0069] The storage component 41 is configured to store various types of data to support the operation of the terminal. The storage component may be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0070] The display component 43 may be an electroluminescent (EL) element, a liquid crystal display or a microdisplay with a similar structure, or a retina direct display or a similar laser scanning display.

[0071] Of course, the computing device may also necessarily include other components, such as an input / output interface, a communication component, etc.

[0072] The input / output interface provides an interface between the processing component and the peripheral interface module, and the peripheral interface module may be an output device, an input device, etc.

[0073] The communication component is configured to facilitate communication between the computing device and other devices in a wired or wireless manner, etc.

[0074] Among them, the computing device may be a physical device or an elastic computing host provided by a cloud computing platform, etc. At this time, the computing device may refer to a cloud server, and the above processing component, storage component, etc. may be basic server resources leased or purchased from a cloud computing platform.

[0075] Embodiment 4, The embodiment of the present application also provides a computer storage medium storing a computer program, and when the computer program is executed by a computer, it can implement the method of the above Figure 1 shown embodiment.

[0076] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments, and will not be elaborated herein.

[0077] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative labor.

[0078] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0079] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, and are not intended to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of each embodiment of the present application.

Claims

1. A method for predicting the state of health of a battery, characterized in that, Including: Obtain a high-dimensional data set during the battery discharge process, and perform normalization processing on the high-dimensional data set to obtain standardized data, where the high-dimensional data set at least includes: current data, voltage data, temperature data, battery capacity data, battery internal resistance data, battery self-discharge rate; Use the Spearman correlation coefficient to analyze the correlation between the standardized data and the battery capacity to obtain a target data set, where the data contained in the target data set are standardized data with a Spearman correlation coefficient greater than a preset threshold; Input the target data set into a width autoencoder, extract the feature nodes and enhancement nodes of the width autoencoder, and obtain an intermediate feature representation; Input the intermediate feature representation into a random configuration network integrated with an intuitionistic fuzzy mechanism to generate a preliminary prediction result of the battery health state.

2. The method according to claim 1, wherein Input the target data set into a width autoencoder, extract the feature nodes and enhancement nodes of the width autoencoder, and obtain an intermediate feature representation, including: Use the mapping layer and random weight matrix of the width autoencoder to convert the target data set into feature nodes to obtain a first-stage feature representation; Expand the first-stage feature representation into the enhancement nodes of the width autoencoder to form a second-stage feature representation; Use the reconstruction layer of the width autoencoder, the second-stage feature representation, and linear weights to reconstruct the target data set to obtain reconstructed data; Minimize the mean square error between the reconstructed data and the target data set to obtain the intermediate feature representation.

3. The method according to claim 2, wherein Input the intermediate feature representation into a random configuration network integrated with an intuitionistic fuzzy mechanism to generate a preliminary prediction result of the battery health state, including: Initialize the random configuration network structure; Input the intermediate feature representation into the initialized random configuration network structure to obtain an initial network output; Use the initial network output and the intuitionistic fuzzy mechanism to iteratively optimize the initialized random configuration network structure to obtain an optimized random configuration network structure; Based on the optimized random configuration network structure, calculate the preliminary prediction result of the battery health state.

4. The method according to claim 3, wherein The method further includes: Combine the physical constraint conditions under the physical information neural network framework into the optimized random configuration network structure to obtain a target model, and set a loss function for the target model, where the loss function includes: a data loss function, a physical equation loss function, and a monotonicity loss function; Based on the loss function and the preliminary prediction result of the battery health state, calculate the loss function value; Use the loss function value to iteratively optimize the target model to obtain an optimized target model; Use the optimized target model to determine the final prediction result of the battery health state.

5. The method according to claim 1, wherein Obtain a high-dimensional data set during the battery discharge process, including: According to a preset battery charge and discharge experiment plan, record the operation data of the battery at different cycle numbers, where the preset battery charge and discharge experiment plan includes a charging strategy and a discharging strategy; Perform preliminary cleaning and preprocessing on the operation data to obtain the high-dimensional data set.

6. The method according to claim 1, wherein Analyze the correlation between the standardized data and the battery capacity using the Spearman correlation coefficient to obtain a target data set, including: Analyze the correlation between the standardized data and the battery capacity using the Spearman correlation coefficient, and calculate the Spearman correlation coefficient corresponding to each standardized data; Construct the target data set using the standardized data with the Spearman correlation coefficient greater than a preset threshold.

7. The method according to claim 4, wherein After obtaining the final prediction result, the method further includes: Obtain the actual result of the battery health state; Calculate an error metric using the actual result and the final prediction result, where the error metric includes: MAPE value and RMSE value; Optimize the optimized target model using the error metric.

8. A battery health state prediction system, characterized in that, Including: An acquisition module, configured to acquire a high-dimensional data set during the battery discharge process, and perform normalization processing on the high-dimensional data set to obtain standardized data, where the high-dimensional data set at least includes: current data, voltage data, temperature data, battery capacity data, battery internal resistance data, battery self-discharge rate; An analysis module, configured to analyze the correlation between the standardized data and the battery capacity using the Spearman correlation coefficient to obtain a target data set, where the data included in the target data set is the standardized data with the Spearman correlation coefficient greater than a preset threshold; An extraction module, configured to input the target data set into a width autoencoder, and extract the feature nodes and enhancement nodes of the width autoencoder to obtain an intermediate feature representation; A generation module, configured to input the intermediate feature representation into a random configuration network integrated with an intuitionistic fuzzy mechanism to generate a preliminary prediction result of the battery health state.

9. A computing device, characterized in that, Including a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement the method according to any one of claims 1 to 7.

10. A computer storage medium, characterized in that, A computer program is stored, and when the computer program is executed by a computer, the method according to any one of claims 1 to 7 is implemented.

Images