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: simulating the operation condition of an energy storage battery through a battery charging and discharging circulation experiment platform, and obtaining the characteristic value of the charging and discharging process of the battery; respectively calculating correlation coefficients corresponding to the characteristic values through a Pearson correlation analysis method and a Spearman correlation analysis method, and selecting the characteristic values of which the correlation coefficients are higher than a preset value as input characteristic values; dividing the input characteristic value into a training set and a test set according to a preset proportion, and training and testing the CNN-LSTM neural network based on the training set and the test set to obtain a CNN-LSTM prediction model after network parameter optimization; and estimating the health state of a battery to be detected through the CNN-LSTM prediction model. Through the scheme, the model setting process can be simplified, the model data complexity is reduced, and the battery health state prediction precision and speed can be improved.

Description

Technical Field

[0001] The present invention belongs to the field of battery technology, and in particular relates to a battery health status prediction method and system. Background Art

[0002] As a power source with high energy density and good economy, lithium-ion batteries have been widely used in energy storage systems, electric vehicles and other industries. With the gradual promotion of applications, current research focuses on the safety and service life of lithium-ion batteries. As the use time increases, lithium batteries will inevitably experience performance degradation, affecting their energy storage capacity and battery safety. In order to avoid safety accidents caused by lithium battery aging, it is necessary to evaluate the health state (SOH) of lithium ions. Due to the complex electrochemical reactions inside lithium batteries and the large differences in operating conditions, the SOH of lithium ions cannot usually be directly measured.

[0003] At present, the SOH estimation methods of lithium-ion batteries mainly include model-based estimation methods and data-driven estimation methods. The model-based estimation method constructs a model that describes the aging behavior of lithium-ion batteries through a series of mathematical methods of algebra and differential equations, and estimates the model parameters online by combining algorithms such as recursive least squares or extended Kalman filtering and their derivative algorithms. However, this model-based estimation method not only has a complex model setting process, but is also easily affected by environmental factors and operating conditions, and has relatively low accuracy under complex conditions.

[0004] Compared with the model-based lithium battery SOH estimation method, the data-driven method usually establishes a mapping relationship between operating data and battery cell SOH to estimate SOH based on the actual data such as voltage, current, operating temperature, and operating time collected during battery operation. The prediction model is basically optimized by combining optimization algorithms and deep learning networks. Although this method improves the prediction accuracy, it is usually not directly deployed on energy storage equipment for online training and testing. Summary of the invention

[0005] In view of this, an embodiment of the present invention provides a battery health state prediction method and system, which are used to solve the problems of complex setting process, low accuracy and inability to detect online of the current battery health state prediction model.

[0006] In a first aspect of an embodiment of the present invention, a method for predicting a battery health state is provided, comprising: The battery charge and discharge cycle experimental platform is used to simulate the operating conditions of the energy storage battery and obtain the characteristic values ​​of the battery charge and discharge process; The correlation coefficients corresponding to the eigenvalues ​​are calculated by Pearson correlation analysis and Spearman correlation analysis respectively, and the eigenvalues ​​with correlation coefficients higher than the preset values ​​are selected as input eigenvalues; The input feature values ​​are divided into a training set and a test set according to a preset ratio, and the CNN-LSTM neural network is trained and tested based on the training set and the test set to obtain a CNN-LSTM prediction model after network parameter optimization; The CNN-LSTM prediction model is used to estimate the health status of the battery to be detected.

[0007] In a second aspect of an embodiment of the present invention, a battery health status prediction system is provided, including: The battery test module is used to simulate the operating conditions of the energy storage battery through the battery charge and discharge cycle experimental platform and obtain the characteristic values ​​of the battery charge and discharge process; A data processing module is used to calculate the correlation coefficients corresponding to the eigenvalues ​​by using the Pearson correlation analysis method and the Spearman correlation analysis method, and select the eigenvalues ​​whose correlation coefficients are higher than the preset values ​​as the input eigenvalues; The model training module is used to divide the input feature values ​​into a training set and a test set according to a preset ratio, and train and test the CNN-LSTM neural network based on the training set and the test set to obtain a CNN-LSTM prediction model after the network parameters are optimized; The model prediction module is used to predict the health status of the battery to be detected through the CNN-LSTM prediction model.

[0008] In a third aspect of an embodiment of the present invention, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method described in the first aspect of the embodiment of the present invention when executing the computer program.

[0009] In a fourth aspect of an embodiment of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method provided in the first aspect of the embodiment of the present invention are implemented.

[0010] In the embodiment of the present invention, the characteristic values ​​of the battery operation process are selected through Pearson and Spearman correlation analysis, and the CNN-LSTM neural network is trained based on the characteristic values. The battery health status is predicted through the trained CNN-LSTM model, which can not only improve the accuracy of battery SOH estimation and simplify the model setting process, but also realize online training and prediction, which is convenient for deployment on embedded systems or devices. The characteristic data is calculated by the correlation coefficient method, and the characteristic data with the highest correlation with SOH is selected, which can reduce data complexity and improve the calculation speed and accuracy in SOH prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0012] Figure 1 A flowchart of a battery health status prediction method provided by an embodiment of the present invention; Figure 2 A schematic diagram of a charge-discharge cycle experiment flow chart provided for one embodiment of the present invention; Figure 3 A schematic diagram of the structure of a battery health status prediction system provided by an embodiment of the present invention; Figure 4 A schematic structural diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0013] In order to make the purpose, features and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0014] It should be understood that the term "including" and other similar expressions in the specification or claims of the present invention and the above-mentioned drawings are intended to cover non-exclusive inclusions, such as a process, method, system, or device including a series of steps or units is not limited to the listed steps or units. In addition, "first" and "second" are used to distinguish different objects, not to describe a specific order.

[0015] See also Figure 1 , a flowchart of a battery health status prediction method provided by an embodiment of the present invention includes: S101, simulating the operating conditions of the energy storage battery through the battery charge and discharge cycle experimental platform, and obtaining characteristic values ​​of the battery charge and discharge process; The battery discharge cycle experiment platform mainly includes three parts: the charging and discharging power supply system, the constant temperature box and the host computer. The charging and discharging power supply system is used to provide constant power supply to charge and discharge the lithium battery, and is equipped with temperature sensors, voltage acquisition, current acquisition and other acquisition equipment. The constant temperature box is used to control the operating environment of the battery cell to maintain a constant temperature and humidity environment. The host computer is used to collect and process data, adjust the cycle test steps, and record lithium battery cycle data.

[0016] The characteristic values ​​are data collected and recorded during the battery charging and discharging process, including average voltage, cycle time, DC internal resistance, initial charging temperature, maximum charging temperature, initial discharging temperature, maximum discharging temperature, charging time, discharging time, and battery cell capacity, etc. The cycle time indicates the total time that the battery cell is cycled in the charge and discharge cycle experimental platform.

[0017] In some embodiments, the characteristic values ​​recorded by the battery charge and discharge cycle test platform experiment are shown in Table 1: Table 1 Experimental record data

[0018] It can be understood that the definition of charging DC internal resistance and discharging DC internal resistance is different from that of conventional DC internal resistance. It is necessary to set the 0s, 30s, and 60s for recording under the two working step conversions of static-charging and static-discharging respectively. The calculation formula is: ; In the formula, R ci Indicates charging i The charging DC internal resistance after seconds, u 0 represents the open circuit voltage of the lithium battery before charging. u ci Indicates lithium battery charging i The voltage after seconds, i ci Indicates lithium battery charging i Actual current per second, discharge DC internal resistance R di The calculation is the same as above.

[0019] Among them, the eigenvalue vector corresponding to each eigenvalue of the battery is constructed to form an eigenvalue matrix, and the target value of the eigenvalue vector is set to the battery cell capacity.

[0020] Construct the battery eigenvalue vectors according to the acquired eigenvalues v i , which is expressed as , where v i,j Indicates i The eigenvalue at time j The numerical value oft Represents the number of cycles. Correspondingly, the target value is set to the cell capacity, and the target vector can be expressed as , where C i Represents the battery capacity value matrix, reflecting the actual capacity of the battery cell. t Indicates the number of cycles, C t Indicates t The cell capacity value of the cycle.

[0021] Therefore, the eigenvalue matrix of the battery cell can be constructed based on the battery eigenvalue vector V n×t : ; In the formula, n Represents the number of eigenvalue types collected, and the eigenvalue matrix is ​​a matrix with a size of Matrix of size.

[0022] S102, respectively calculating the correlation coefficients corresponding to the eigenvalues ​​by using the Pearson correlation analysis method and the Spearman correlation analysis method, and selecting the eigenvalues ​​whose correlation coefficients are higher than the preset values ​​as the input eigenvalues; Pearson correlation analysis is generally a statistical method used to measure the degree of linear correlation between two variables. It calculates the covariance and standard deviation of the two variables to obtain a value between -1 and 1 to indicate the degree of correlation of the variables. Spearman correlation analysis is a non-parametric statistical method used to evaluate the monotonic relationship between two random variables. It measures the correlation by calculating the rank of the variables. By using these two correlation analysis methods to calculate the correlation between each eigenvalue and the cell capacity (i.e., the target value), the eigenvalues ​​can be screened and the model characteristic parameters can be simplified.

[0023] Among them, the Pearson correlation coefficient and Spearman correlation coefficient corresponding to each eigenvalue are calculated according to formulas (1) and (2): ; (1) ; (2) In the formula, represents the Pearson correlation coefficient, represents the Spearman correlation coefficient, x i Indicates i The characteristic value at the moment, y i Indicates i The cell capacity at a given moment, , Represent the characteristic value and the average value of the cell capacity, n represents the number of eigenvalue cycles, d Represents eigenvalue x i , y i The difference in ranking order after they are arranged in sequence.

[0024] In some embodiments, the correlation coefficient between the calculated eigenvalue and the cell charging capacity is shown in Table 2. A correlation coefficient between 0.8 and 1 generally indicates a strong correlation between the data, and eigenvalues ​​with Pearson coefficient and Spearman coefficient higher than 0.8 can be selected as inputs to the subsequent CNN-LSTM battery SOH prediction network.

[0025] Table 2 Correlation coefficient

[0026] According to the eigenvalue correlation coefficients calculated in Table 2, the eigenvalues ​​with higher absolute values ​​of the Pearson coefficient and the Spearman coefficient can be selected as the input of the subsequent prediction model.

[0027] S103, dividing the input feature values ​​into a training set and a test set according to a preset ratio, and training and testing the CNN-LSTM neural network based on the training set and the test set to obtain a CNN-LSTM prediction model after network parameter optimization; According to the preset ratio, the eigenvalue matrix is ​​divided into training set and test set in a certain ratio, which can be expressed as: ; ; ; ; In the formula, V train , V test denote the training set and the test set respectively, α It indicates the ratio of the training set to the test set, that is, the preset ratio. The preset ratio is generally between 0.7 and 0.9. t 1. t 2 represent the matrix lengths of the training set and the test set respectively, and there are t = t 1+ t 2.

[0028] The CNN-LSTM neural network is a hybrid neural network model that combines a convolutional neural network (CNN) and a long short-term memory network (LSTM). It extracts feature information of data through CNN, and then uses LSTM to process the feature information to achieve time series prediction. The superior performance of the CNN network in data feature recognition, combined with the LSTM network with long-term memory characteristics to form a CNN-LSTM network, can improve the prediction accuracy of battery charging capacity.

[0029] The CNN network consists of an input layer, a convolution layer, and a pooling layer. The core part is the convolution layer, which uses the convolution kernel to perform a convolution operation on the input. The calculation formula is: , where σ represents the activation function, and the Sigmoid function is selected as the activation function. w represents the convolution kernel, x represents the feature vector, b represents the offset, and * represents the convolution operation.

[0030] The LSTM network can memorize the characteristics of time series data and filter time series data by introducing gating units such as input gate, output gate, and forget gate. σ represents the sigmoid neural layer (the neural network parameter is W i , b i ), tanh represents the tanh neural layer (the neural network parameters are W c , b c ), the LSTM network update formula is: ; ; ; ; ; ; In the formula, represents the Hadamard product, f t Forget gate, controls the degree of memory, i t Is the input gate, which controls the input to the memory unit C t The degree of influence o t is the output gate, controlling the input ratio, Indicates that through a tanh The neural layer gets a new candidate LSTM state, , Respectively represent the LSTM state at the current moment and the previous moment, , Represent the LSTM hidden states at the current moment and the previous moment respectively.

[0031] Optionally, the training set is passed forward layer by layer in each network layer of the CNN-LSTM neural network, and the network parameters are iteratively updated in combination with the Adam optimization algorithm and the cross entropy loss function until the maximum number of iterations is reached, thereby obtaining a CNN-LSTM prediction model after the network parameters are optimized.

[0032] Combining the Adam optimization algorithm and the cross entropy loss function, the neural network parameters are iteratively updated until the maximum number of iterations is reached, and the optimal neural network parameters can be obtained.

[0033] Optionally, the test set is brought into the trained CNN-LSTM model for calculation to obtain the corresponding mean absolute error (MAE) and the root mean square error (RMSE) reflecting the degree of data dispersion. The calculation formula is: ; ; In the formula, n is the number of predicted values, y i is the predicted value, is the actual value, is the average of the actual values; According to the MAE and RMSE values, judge whether the CNN-LSTM model meets the requirements. When MAE and RMSE are closer to 0, the neural network prediction results are better. Generally, when MAE and RMSE are lower than 5% of the predicted feature value, the CNN-LSTM model prediction is basically accurate, otherwise it needs to be retrained. S104, estimating the health status of the battery to be detected by using the CNN-LSTM prediction model.

[0034] Collect the characteristic data of the battery to be tested and use the CNN-LSTM prediction model to estimate the health status of the battery to be tested.

[0035] The current battery charging capacity is estimated by the CNN-LSTM prediction model, and the corresponding battery health status value is calculated according to formula (3): (3) In the formula, SOH represents the battery health status value, C Indicates the battery charging capacity in the current cycle. C max indicates the maximum charge capacity of the battery.

[0036] For example, a lithium-ion battery discharge cycle test platform is used to perform cycle tests on lithium batteries and record lithium battery cycle data. The test battery object uses the 280Ah 3.2V lithium iron phosphate battery for energy storage, which is the mainstream of energy storage systems. Due to the large size of the battery and uneven temperature distribution, temperature sensors are set at the negative ear, side, and bottom of the battery in the charge and discharge cycle experiment. The constant temperature box environment is set to a constant 25°C and the charge and discharge power is set to a constant 448W. The steps of the charge and discharge cycle experiment are as follows: Figure 2 As shown, the battery is discharged to 2.5V at constant power, and then charged to 3.65V at constant power after 30 minutes of rest. After 30 minutes of rest, it is discharged to 2.5V at constant power. When the charging capacity is less than 224Ah or the number of cycles is reached, the cyclic charge and discharge experiment is terminated. Otherwise, it is charged at constant power after 30 minutes of rest. The characteristic data of 425 charge and discharge cycles of two single cells C1 and C2 of the same model and batch are collected.

[0037] According to the recorded characteristic data, calculate its characteristic value and battery capacity C Usually, a correlation coefficient between 0.8 and 1 indicates a strong correlation between the data. Therefore, we can select 6 eigenvalues ​​with higher Pearson and Spearman coefficients. V mean , t c , t d , R c30 , R d0 , N As the input of the subsequent CNN-LSTM battery SOH prediction network.

[0038] The above eigenvalues ​​are used as inputs to predict the charging capacity of batteries C1 and C2 using the CNN-LSTM network. The first 350 sets of data are set as training sets, and the last 75 sets of data are set as test sets. The convolution kernel size of the convolution layer in the CNN-LSTM network is 3x3, the initial number of convolution kernels is 5, the training batch is 4, and the learning rate is 0.001. The network contains 1 LSTM layer, the initial number of hidden layer nodes is 50, and the learning rate is 0.005. Among them, the number of convolution kernels is constrained to [1, 16], the learning rate is constrained to [0.001, 0.02], and the hidden layer node constraint is [20, 100].

[0039] In order to evaluate the effectiveness and stability of the trained CNN-LSTM model, the mean absolute error (MAE) reflecting the consistency between data and the root mean square error (RMSE) reflecting the degree of data dispersion were used to evaluate its prediction results. The CNN-LSTM network predicted the charging capacity of batteries C1 and C2, and the absolute average errors were 0.5334Ah (error of about 2%) and 0.6299Ah (error of about 2.5%), respectively, verifying the effectiveness and accuracy of the CNN-LSTM battery charging capacity prediction network. Therefore, the CNN-LSTM model can be used to predict battery SOH.

[0040] In this embodiment, the characteristic data is processed by Pearson and Spearman correlation analysis to select the characteristic value with the highest correlation with SOH, so as to reduce the data complexity of the model and improve the accuracy and efficiency of SOH prediction. The health status of energy storage lithium-ion batteries is estimated based on the CNN-LSTM model, which can realize online detection. It not only has a small amount of calculation and high accuracy, but also has a simple model setting process, which is convenient for deployment on embedded systems or mobile devices.

[0041] It should be understood that the serial numbers of the steps in the above embodiments do not imply a sequence of execution, and the execution sequence of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0042] Figure 3 A schematic diagram of a battery health status prediction system provided by an embodiment of the present invention, the system comprising: The battery testing module 310 is used to simulate the operating conditions of the energy storage battery through the battery charge and discharge cycle experimental platform and obtain the characteristic values ​​of the battery charge and discharge process; Wherein, the obtaining of characteristic values ​​of the battery charging and discharging process includes: The eigenvalue vectors corresponding to each eigenvalue of the battery are constructed and an eigenvalue matrix is ​​formed, and the target value of the eigenvalue vector is set as the cell capacity.

[0043] The data processing module 320 is used to calculate the correlation coefficients corresponding to the eigenvalues ​​by using the Pearson correlation analysis method and the Spearman correlation analysis method, and select the eigenvalues ​​whose correlation coefficients are higher than the preset values ​​as the input eigenvalues; Specifically, the Pearson correlation coefficient and Spearman correlation coefficient corresponding to each eigenvalue are calculated according to formulas (1) and (2): ; (1) ; (2) In the formula, represents the Pearson correlation coefficient, represents the Spearman correlation coefficient, x i Indicates i The characteristic value at the moment, y i Indicates i The cell capacity at a given moment, , Represent the characteristic value and the average value of the cell capacity, n represents the number of eigenvalue cycles, d Represents eigenvalue x i , y i The difference in ranking order after they are arranged in sequence.

[0044] The model training module 330 is used to divide the input feature values ​​into a training set and a test set according to a preset ratio, and train and test the CNN-LSTM neural network based on the training set and the test set to obtain a CNN-LSTM prediction model after the network parameters are optimized; Optionally, the training set is passed forward layer by layer in each network layer of the CNN-LSTM neural network, and the network parameters are iteratively updated in combination with the Adam optimization algorithm and the cross entropy loss function until the maximum number of iterations is reached, thereby obtaining a CNN-LSTM prediction model after the network parameters are optimized.

[0045] The model prediction module 340 is used to predict the health status of the battery to be detected by using the CNN-LSTM prediction model.

[0046] The current battery charging capacity is estimated by the CNN-LSTM prediction model, and the corresponding battery health status value is calculated according to formula (3): (3) In the formula, SOH represents the battery health status value, C represents the battery charging capacity in the current cycle, C max Indicates the maximum charge capacity of the battery.

[0047] In some embodiments, the battery health status prediction system further includes a data acquisition module for acquiring characteristic data of the battery to be tested so as to estimate the battery health status through the model prediction module.

[0048] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems and modules can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here.

[0049] Figure 4 1 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. The electronic device is used for estimating the health status of a battery. Figure 4 As shown, the electronic device 4 of this embodiment includes: a memory 410, a processor 420 and a system bus 430, wherein the memory 410 includes an executable program 4101 stored thereon, and those skilled in the art can understand that Figure 4 The electronic device structure shown in the figure does not constitute a limitation of the electronic device, and may include more or less components than shown in the figure, or combine certain components, or arrange the components differently.

[0050] Combine the following Figure 4 A detailed introduction to the various components of electronic equipment: The memory 410 can be used to store software programs and modules. The processor 420 executes various functional applications and data processing of the electronic device by running the software programs and modules stored in the memory 410. The memory 410 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area may store data created according to the use of the electronic device (such as cache data), etc. In addition, the memory 410 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0051] The memory 410 includes an executable program 4101 of the network request method, and the executable program 4101 can be divided into one or more modules / units, which are stored in the memory 410 and executed by the processor 420 to achieve battery health status prediction, etc. The one or more modules / units can be a series of computer program instruction segments that can complete specific functions, and the instruction segments are used to describe the execution process of the computer program 4101 in the electronic device 4. For example, the computer program 4101 can be divided into functional modules such as a battery testing module, a data processing module, a model training module, and a model prediction module.

[0052] The processor 420 is the control center of the electronic device. It uses various interfaces and lines to connect various parts of the entire electronic device. It executes various functions of the electronic device and processes data by running or executing software programs and / or modules stored in the memory 410, and calling data stored in the memory 410, so as to monitor the overall status of the electronic device. Optionally, the processor 420 may include one or more processing units; preferably, the processor 420 may integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, application programs, etc., and the modem processor mainly processes wireless communications. It is understandable that the above-mentioned modem processor may not be integrated into the processor 420.

[0053] The system bus 430 is used to connect the various functional components inside the computer, and can transmit data information, address information, and control information. Its types can be, for example, PCI bus, ISA bus, CAN bus, etc. The instructions of the processor 420 are transmitted to the memory 410 through the bus, and the memory 410 feeds back data to the processor 420. The system bus 430 is responsible for the data and instruction exchange between the processor 420 and the memory 410. Of course, the system bus 430 can also be connected to other devices, such as network interfaces, display devices, etc.

[0054] In the embodiment of the present invention, the executable program executed by the processor 420 included in the electronic device includes: The battery charge and discharge cycle experimental platform is used to simulate the operating conditions of the energy storage battery and obtain the characteristic values ​​of the battery charge and discharge process; The correlation coefficients corresponding to the eigenvalues ​​are calculated by Pearson correlation analysis and Spearman correlation analysis respectively, and the eigenvalues ​​with correlation coefficients higher than the preset values ​​are selected as input eigenvalues; The input feature values ​​are divided into a training set and a test set according to a preset ratio, and the CNN-LSTM neural network is trained and tested based on the training set and the test set to obtain a CNN-LSTM prediction model after network parameter optimization; The CNN-LSTM prediction model is used to estimate the health status of the battery to be detected.

[0055] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and modules described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0056] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0057] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting battery health status, characterized in that: include: The battery charge and discharge cycle experimental platform is used to simulate the operating conditions of the energy storage battery and obtain the characteristic values ​​of the battery charge and discharge process; The correlation coefficients corresponding to the eigenvalues ​​are calculated by Pearson correlation analysis and Spearman correlation analysis respectively, and the eigenvalues ​​with correlation coefficients higher than the preset values ​​are selected as input eigenvalues; The input feature values ​​are divided into a training set and a test set according to a preset ratio, and the CNN-LSTM neural network is trained and tested based on the training set and the test set to obtain a CNN-LSTM prediction model after network parameter optimization; The CNN-LSTM prediction model is used to estimate the health status of the battery to be detected.

2. The method according to claim 1, characterized in that The obtaining of characteristic values ​​of the battery charging and discharging process comprises: The eigenvalue vectors corresponding to each eigenvalue of the battery are constructed to form an eigenvalue matrix, and the target value of the eigenvalue vector is set as the cell capacity.

3. The method according to claim 1, characterized in that The calculation of the correlation coefficients corresponding to the eigenvalues ​​by the Pearson correlation analysis method and the Spearman correlation analysis method respectively includes: According to formulas (1) and (2), the Pearson correlation coefficient and Spearman correlation coefficient corresponding to each eigenvalue are calculated respectively: ； （1） ；（2） In the formula, represents the Pearson correlation coefficient, represents the Spearman correlation coefficient, x i Indicates i The characteristic value at the moment, y i Indicates i The cell capacity at a given moment, , Represent the characteristic value and the average value of the cell capacity, n represents the number of eigenvalue cycles, d Represents eigenvalue x i , y i The difference in ranking order after they are arranged in sequence.

4. The method according to claim 1, characterized in that The input feature values ​​are divided into a training set and a test set according to a preset ratio, and the CNN-LSTM neural network is trained and tested based on the training set and the test set to obtain a CNN-LSTM prediction model after the network parameters are optimized, including: The training set is passed forward layer by layer in each network layer of the CNN-LSTM neural network. Combined with the Adam optimization algorithm and the cross entropy loss function, the network parameters are iteratively updated until the maximum number of iterations is reached, and the CNN-LSTM prediction model with optimized network parameters is obtained.

5. The method according to claim 1, characterized in that The estimating the health status of the battery to be detected by using the CNN-LSTM prediction model includes: The current battery charging capacity is estimated by the CNN-LSTM prediction model, and the corresponding battery health status value is calculated according to formula (3): （3） In the formula, SOH represents the battery health status value, C Indicates the battery charging capacity in the current cycle. C max Indicates the maximum charge capacity of the battery.

6. A battery health status prediction system, characterized in that: include: The battery test module is used to simulate the operating conditions of the energy storage battery through the battery charge and discharge cycle experimental platform and obtain the characteristic values ​​of the battery charge and discharge process; A data processing module is used to calculate the correlation coefficients corresponding to the eigenvalues ​​by using the Pearson correlation analysis method and the Spearman correlation analysis method, and select the eigenvalues ​​whose correlation coefficients are higher than the preset values ​​as the input eigenvalues; The model training module is used to divide the input feature values ​​into a training set and a test set according to a preset ratio, and train and test the CNN-LSTM neural network based on the training set and the test set to obtain a CNN-LSTM prediction model after the network parameters are optimized; The model prediction module is used to predict the health status of the battery to be detected through the CNN-LSTM prediction model.

7. The system according to claim 6, characterized in that The obtaining of characteristic values ​​of the battery charging and discharging process comprises: The eigenvalue vectors corresponding to each eigenvalue of the battery are constructed and an eigenvalue matrix is ​​formed, and the target value of the eigenvalue vector is set as the cell capacity.

8. The system according to claim 6, characterized in that The calculation of the correlation coefficients corresponding to the eigenvalues ​​by the Pearson correlation analysis method and the Spearman correlation analysis method respectively includes: According to formulas (1) and (2), the Pearson correlation coefficient and Spearman correlation coefficient corresponding to each eigenvalue are calculated respectively: ； （1） ；（2） In the formula, represents the Pearson correlation coefficient, represents the Spearman correlation coefficient, x i Indicates i The characteristic value at the moment, y i Indicates i The cell capacity at a given moment, , Represent the characteristic value and the average value of the cell capacity, n represents the number of eigenvalue cycles, d Represents eigenvalue x i , y i The difference in ranking order after they are arranged in sequence.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of a battery health status prediction method as described in any one of claims 1 to 5 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed, the steps of a battery health status prediction method as described in any one of claims 1 to 5 are implemented.

Images