A lithium ion battery temperature estimation method and system based on electrochemical impedance spectroscopy

Abstract

The application discloses a lithium ion battery temperature estimation method and system based on electrochemical impedance spectroscopy. The method first performs electrochemical impedance spectroscopy testing on the same type of lithium ion battery under different states of charge, different charge and discharge cycle numbers, and different temperature conditions to obtain electrochemical impedance spectroscopy data of the battery. Based on electrochemical impedance spectroscopy analysis, a characteristic frequency with low coupling degree to the state of charge and the charge and discharge cycle number and sensitive to temperature is selected, and corresponding characteristic impedance parameters are extracted. Combined with correlation analysis, electrochemical impedance spectroscopy characteristic variables suitable for temperature estimation are determined. The global optimization of hyperparameters of the extreme gradient boosting algorithm is performed by using the Canthon optimization algorithm, and a Canthon optimization-extreme gradient boosting temperature estimation model is trained. The battery electrochemical impedance spectroscopy characteristics are input into the model, and the temperature estimation result of the lithium ion battery is output. The method is suitable for electric vehicles, energy storage systems, portable electronic devices, and the like.

Description

Technical Field

[0001] This invention belongs to the field of lithium-ion battery temperature estimation technology, and particularly relates to a lithium-ion battery temperature estimation method and system based on electrochemical impedance spectroscopy. Background Technology

[0002] Lithium-ion batteries are widely used in portable electronics, power batteries, and large-scale energy storage due to their high energy density, long cycle life, and high energy conversion efficiency. Battery temperature is a key parameter affecting its safety, lifespan, and performance. Excessively high temperatures accelerate electrochemical side reactions, cause electrolyte decomposition, and may lead to thermal runaway, while excessively low temperatures significantly reduce the battery's power output and cycle performance.

[0003] Existing methods for obtaining battery temperature mainly include direct measurement methods, model-based temperature estimation methods, and data-driven temperature estimation methods. Direct measurement methods (such as external NTC, embedded thermocouples, or fiber Bragg gratings) can directly obtain temperature values, but embedded solutions increase manufacturing complexity and may damage the battery structure, while external solutions are difficult to accurately reflect the battery temperature field. Model-based temperature estimation methods have complex models, are difficult to identify parameters, and are sensitive to changes in operating conditions. Model verification and calibration are labor-intensive, making it difficult to achieve high-precision online solutions directly in resource-constrained battery management systems. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for estimating the temperature of lithium-ion batteries based on electrochemical impedance spectroscopy, in order to solve the problems mentioned in the background art.

[0005] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a method for estimating the temperature of lithium-ion batteries based on electrochemical impedance spectroscopy, comprising the following steps:

[0006] Step 1: Measure the EIS data of the same type of lithium-ion battery at different SOCs, different charge-discharge cycles, and different temperatures;

[0007] Step 2: Perform feature analysis on the EIS data obtained in Step 1, and select one or more feature frequencies that meet the requirements of correlation with SOC, charge-discharge cycle number and temperature through statistical and correlation analysis.

[0008] Step 3: Extract the electrochemical impedance spectroscopy data obtained in Step 1 according to the characteristic frequencies determined in Step 2, and select characteristic variables based on the extraction results and correlation analysis;

[0009] Step 4: Preprocess the dataset obtained in Step 3, and use dung beetle optimization to globally optimize the hyperparameters of the XGBoost model and train the model. Finally, input the EIS features of the test set into the model to output the battery temperature estimate.

[0010] Based on the same inventive concept, the present invention also provides a lithium-ion battery temperature estimation system based on electrochemical impedance spectroscopy, comprising:

[0011] The first module is used to perform EIS testing on lithium-ion batteries of the same model at different SOCs, different charge-discharge cycles, and different temperatures to obtain the corresponding electrochemical impedance spectroscopy data.

[0012] The second module, based on the battery electrochemical impedance spectroscopy, selects one or more characteristic frequencies that meet the requirements for correlation with SOC, charge-discharge cycle number and temperature through statistical and correlation analysis.

[0013] The third module, in conjunction with the characteristic frequencies determined in the second module, extracts the EIS characteristic parameters at the corresponding frequency points and, in conjunction with correlation analysis, determines the characteristic variables suitable for temperature estimation.

[0014] The fourth module uses DBO to adaptively optimize the hyperparameters of the XGBoost model globally and completes model training based on the training set data; the EIS feature variables of the test set are input into the optimized and trained model to output the temperature estimation results of the lithium-ion battery.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) The present invention achieves temperature estimation based on EIS features, without the need to embed sensors inside the battery, avoiding the damage to the battery structure and the increase in manufacturing complexity caused by traditional embedded temperature measurement methods. At the same time, it overcomes the defect that external temperature measurement is difficult to accurately reflect the internal temperature of the cell, and realizes safe and non-invasive internal temperature estimation. (2) The present invention does not require complex mechanistic thermal models or high-dimensional parameter identification. It only relies on the function mapping between EIS features and temperature to achieve temperature estimation. The computational load is small, and it is easy to integrate into the Battery Management System (BMS) to realize online temperature monitoring. Attached Figure Description

[0016] Figure 1 This is a flowchart of the lithium-ion battery temperature estimation method based on electrochemical impedance spectroscopy of the present invention.

[0017] Figure 2 This is a schematic diagram showing the change of electrochemical impedance spectroscopy with frequency at different SOCs.

[0018] Figure 3 This is a schematic diagram showing the change of electrochemical impedance spectroscopy with frequency at different charging cycles.

[0019] Figure 4 This is a schematic diagram showing the change of electrochemical impedance spectroscopy with frequency at different temperatures.

[0020] Figure 5 This is a schematic diagram showing the variation of the real part of the impedance with frequency under different SOCs.

[0021] Figure 6 This is a schematic diagram showing the change of the imaginary part of impedance with frequency under different SOCs.

[0022] Figure 7 This is a schematic diagram showing the change of impedance magnitude with frequency under different SOCs.

[0023] Figure 8 This is a schematic diagram showing the impedance phase variation with frequency under different SOCs.

[0024] Figure 9 This is a schematic diagram showing the change of the real part of the impedance with frequency for different number of cycles.

[0025] Figure 10 This is a schematic diagram showing the change of the imaginary part of the impedance with frequency for different number of cycles.

[0026] Figure 11 This is a schematic diagram showing the change of impedance magnitude with frequency under different number of cycles.

[0027] Figure 12 This is a schematic diagram showing the change of impedance phase with frequency under different number of cycles.

[0028] Figure 13 This is a schematic diagram showing how the real part of the impedance changes with frequency at different temperatures.

[0029] Figure 14 This is a schematic diagram showing how the imaginary part of the impedance changes with frequency at different temperatures.

[0030] Figure 15 This is a schematic diagram showing how the impedance modulus changes with frequency at different temperatures.

[0031] Figure 16 This is a schematic diagram showing the change of impedance phase with frequency at different temperatures.

[0032] Figure 17 The correlation between EIS parameters and SOC.

[0033] Figure 18 The correlation between EIS parameters and the number of charging cycles.

[0034] Figure 19 The correlation between EIS parameters and temperature.

[0035] Figure 20 The convergence curve of the fitness function value of the DBO optimization algorithm. Detailed Implementation

[0036] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0037] like Figure 1 As shown, the present invention provides a method for estimating the temperature of a lithium-ion battery based on electrochemical impedance spectroscopy, comprising the following steps:

[0038] Step 1: Measure the electrochemical impedance spectroscopy (EIS) data of the same type of lithium-ion battery at different states of charge (SOC), different charge-discharge cycles, and different temperatures;

[0039] Step 2: Perform feature analysis on the EIS data obtained in Step 1, and select one or more characteristic frequencies that are low in coupling with SOC and charge / discharge cycle count and sensitive to temperature through statistical and correlation analysis.

[0040] Step 3: Extract the electrochemical impedance spectroscopy data obtained in Step 1 according to the characteristic frequencies determined in Step 2, and select characteristic variables based on the extraction results and correlation analysis;

[0041] Step 4: Preprocess the dataset obtained in Step 3, and use Dung Beetle Optimizer (DBO) to globally optimize the hyperparameters of the XGBoost model and train the model. Input the EIS feature variables of the test set into the optimized and trained model, and output the temperature estimation results of the lithium-ion battery.

[0042] Furthermore, in this embodiment, step one specifically includes:

[0043] Step 1-1: Take batteries of the same model that have been cycled 0, 100, 200, 300, and 400 times, charge the batteries to 100% SOC at 25°C using the CC-CV method, and let them stand for 4 hours.

[0044] Steps 1-2: Discharge the battery with a current of 0.5C for 12 minutes to discharge the battery to 90% SOC. Then, place the battery in a high and low temperature test chamber set to 25℃ and let it stand for 8 hours.

[0045] Steps 1-3: Perform EIS testing on the battery using an electrochemical workstation;

[0046] Steps 1-4: Increase the temperature of the high and low temperature test chamber by 5°C and let it stand for another 8 hours.

[0047] Steps 1-5, repeat steps 1-3-1-4, until the temperature reaches 50℃;

[0048] Steps 1-6: Reset the high and low temperature test chamber to 25℃ and let it stand for 8 hours;

[0049] Steps 1-7: Discharge the battery with a current of 0.5C for 24 minutes, then let it stand for 8 hours;

[0050] Steps 1-8, repeat steps 1-3 through 1-7, until the loop reaches 10 times.

[0051] Furthermore, in this embodiment, step two specifically involves:

[0052] The PCC method is used to quantitatively analyze the correlation between data from various characteristic points and battery SOC, temperature, and charge-discharge cycle count. The PCC formula can be defined as:

[0053]

[0054] in, Let (X, Y) be the covariance between variables. PCC is the product of the standard deviations of variables (X, Y). The value of PCC is between [-1, 1]. When its value is close to ±1, there is a strong positive and negative correlation between variables X and Y. The smaller the absolute value, the weaker the correlation between the two variables. The correlation between impedance parameters and SOC, charge-discharge cycle number and temperature at each frequency point is quantified by PCC. The characteristic frequencies with the imaginary part of impedance and phase as follows are selected: |PCC| ≤ 0.65 with SOC, |PCC| ≤ 0.65 with SOH (charge-discharge cycle number), and |PCC| ≥ 0.94 with temperature.

[0055] Furthermore, in this embodiment, step three specifically involves:

[0056] The electrochemical impedance spectroscopy data obtained in step one are extracted according to the characteristic frequencies determined in step two. Combined with correlation analysis, the imaginary part of the battery impedance and the phase value measured at the characteristic frequencies are selected as characteristic variables.

[0057] Furthermore, in this embodiment, step four specifically involves:

[0058] The dataset obtained in step three is divided into a training set and a test set according to a predetermined ratio. The training set is further divided into a training subset and a validation subset in each round through k-fold cross-validation (optional, k=5) for hyperparameter search and model selection.

[0059] The training and validation subsets are input into the Extreme Gradient Boosting (XGBoost) regressor. Based on this, the Dung Beetle Optimizer (DBO) algorithm is introduced to adaptively optimize the key hyperparameters of XGBoost (including learning rate, maximum tree depth, subsample rate, column sample rate, and number of base learners n_estimators, etc.). The optimization objective is the temperature estimation error (such as RMSE or MAE) on the validation set.

[0060] The optimal hyperparameters obtained through DBO search were used to train the final XGBoost model, forming the DBO-XGBoost temperature estimation model.

[0061] The reserved test set is input into the optimized and trained DBO-XGBoost model to evaluate and output the temperature estimation results and performance indicators of the lithium-ion battery.

[0062] This invention also provides a lithium-ion battery temperature estimation system based on electrochemical impedance spectroscopy, used to implement the above-mentioned lithium-ion battery temperature estimation method. The system includes:

[0063] The first module is used to perform EIS testing on lithium-ion batteries of the same model at different SOCs, different charge-discharge cycles, and different temperatures to obtain the corresponding electrochemical impedance spectroscopy data.

[0064] The second module, based on the battery electrochemical impedance spectroscopy, selects one or more characteristic frequencies that meet the requirements for correlation with SOC, charge-discharge cycle number and temperature through statistical and correlation analysis.

[0065] The third module, in conjunction with the characteristic frequencies determined in the second module, extracts the EIS characteristic parameters at the corresponding frequency points and, in conjunction with correlation analysis, determines the characteristic variables suitable for temperature estimation.

[0066] The fourth module uses DBO to adaptively optimize the hyperparameters of the XGBoost model globally and completes model training based on the training set data; the EIS feature variables of the test set are input into the optimized and trained model to output the temperature estimation results of the lithium-ion battery.

[0067] The specific implementation method of the above module is the same as the aforementioned lithium-ion battery temperature estimation method, and will not be repeated here.

[0068] The present invention will be specifically described below using a ternary lithium battery as an example.

[0069] Example

[0070] Cylindrical ternary lithium batteries of model INR18650-26E were selected as the experimental subject. Batteries of the same model with 0, 100, 200, 300, and 400 cycles were subjected to EIS testing. The specific procedure is as follows:

[0071] 1) Charge the battery to 100% SOC using CC-CV method at 25℃ and let it rest for 8 hours;

[0072] 2) Discharge the battery with a current of 0.5C for 12 minutes, then place the battery in a high and low temperature test chamber set at 25℃ for 8 hours;

[0073] 3) Perform EIS testing on the battery using an electrochemical workstation;

[0074] 4) Increase the temperature of the high and low temperature test chamber by 5°C and let it stand for 8 hours;

[0075] 5) Repeat steps 3)-4) until the temperature reaches 50℃;

[0076] 6) Reset the high and low temperature test chamber to 25℃ and let it stand for 8 hours;

[0077] 7) Discharge the battery at a current of 0.5C for 24 minutes, then let it rest for 8 hours;

[0078] The schematic diagrams of the electrochemical impedance spectra obtained in the experiment as a function of frequency at different SOCs, charging cycle numbers, and temperatures are shown below. Figure 2 , Figure 3 , Figure 4 As shown. Analysis Figures 2-4 It is evident that the battery EIS curve is significantly affected by the number of charge-discharge cycles and temperature, while being less affected by SOC. Furthermore, the trends of change in different regions of the curve vary due to battery aging and temperature influences. Therefore, further analysis of the battery EIS data is needed to determine its correlation with SOC, charge-discharge cycle count, and temperature, and to extract characteristic variables that can effectively characterize battery temperature. Figures 5-8 It can be seen that the imaginary part and phase of the battery impedance are less affected by the battery's state of charge (SOC), and the real part and magnitude curves of the battery impedance do not shift significantly under different SOCs within the frequency range of 10-100kHz; below 10Hz, the real part and magnitude curves of the battery impedance show a clear upward trend as the battery SOC decreases. Analysis Figures 9-12 It can be seen that at most test frequencies, the real part and magnitude of the battery impedance increase significantly with the increase of the number of cycles; in contrast, the imaginary part and phase of the impedance are not sensitive to changes in the number of cycles. Analysis Figures 13-16It can be seen that as the temperature increases, both the real part and the magnitude of the impedance of the battery at each test frequency show a significant decreasing trend; in the frequency range of 1–10 kHz, the imaginary part of the impedance shows a slight decrease; while the impedance phase angle increases significantly with temperature, showing a positive correlation with temperature. Correlation analysis was performed on EIS data at 12 test frequency points in the 1–3 kHz range, and the results are as follows: Figures 17-19 As shown. By Figure 17 As shown, in the frequency range of 300-1kHz, the PCC of the imaginary part and phase of the battery impedance with SOC is generally below 0.65. In the frequency range above 1kHz, the correlation between the imaginary part and phase of the battery impedance and SOC increases. Figure 18 It can be seen that at all frequency points, the average PCC values ​​of the real part and magnitude of the battery impedance with respect to the State of Charge (SOC) are 0.911 and 0.910, respectively, both showing a strong correlation with the SOC. Furthermore, in the frequency range above 10Hz, the absolute values ​​of the PCC values ​​of the imaginary part and phase of the battery impedance with respect to the SOC are generally below 0.65. Figure 19 It can be seen that among all 12 selected frequencies, almost all four parameters of the battery EIS exhibited a PCC greater than 0.8 with temperature. In the high-frequency range above 100Hz, the PCC of each parameter with temperature was above 0.94, showing a strong correlation. After comprehensive analysis, the imaginary part of impedance and phase values ​​at two measurement frequency points of 794.3Hz and 398.1Hz in the 300-1kHz range were selected as characteristic variables to characterize battery temperature.

[0079] Based on the PCC results, the imaginary part of the battery impedance and the phase value measured at 794.3Hz and 398.1Hz were selected as the eigenvectors [Z]. real1 Z imag1 Z real2 Z imag2 ]. Among them, Z real1 Z imag1 and Z real2 Z imag2 The values ​​are the real and imaginary parts of the battery impedance at test frequencies of 794.3 Hz and 398.1 Hz, respectively.

[0080] First, after feature filtering of the collected battery EIS data, a dataset D={( , )}, where x i Let [Z] be the feature vector of the i-th sample. real1 Z imag1 Z real2 Z imag2 ]; This is the actual temperature value of the battery. In the XGBoost algorithm, each regression tree... The mapping relationship between feature vectors and battery temperature will be constructed, and the ensemble model will be formed by the linear weighting of M regression trees. It outputs the model's temperature estimate. Its mathematical model can be expressed as:

[0081]

[0082] In the formula, For the first A regression tree, The parameters are defined as follows. During iterative training, the model performance is improved by increasing the error between the weak classifier's fit to the previous training model result and the sample data, thereby enhancing the accuracy of the XGBoost algorithm in estimating battery temperature. The objective function of the model is shown below:

[0083]

[0084] In the formula, It is a loss function used to quantify the battery temperature estimate of the model. and the actual battery temperature in the sample The error; This is the regularization term, also known as the penalty function. Let be the m-th regression tree; γ is used to quantify the complexity of each leaf in the regression tree; T is the number of leaves in a tree; λ is a parameter that measures the penalty. These are the weights (output values) of the leaf nodes. To minimize the model's objective function, new trees can be added. When m trees are generated, the model's temperature estimate is:

[0085]

[0086] In the formula; This is the temperature estimate from the model in the previous iteration. This is the m-th regression tree added in this iteration. At this point, the model's objective function is:

[0087]

[0088] To quickly optimize the model's objective function, a second-order approximation transformation of the objective function can be performed using Taylor series expansion:

[0089]

[0090] In the formula, , These represent the first and second derivatives of the loss function, respectively. When the number of regression trees in the model, i.e., the number of iterations, reaches the set maximum number of iterations (Max), the final XGBoost model function is obtained:

[0091]

[0092] In the formula, This is the battery temperature estimate output by the model after the last round of training iterations.

[0093] The DBO algorithm is introduced into the hyperparameter optimization of the XGBoost algorithm. The root mean square error (RMSE) of temperature estimation is used as the fitness function of the DBO algorithm, and the hyperparameters in the XGBoost algorithm are the variables to be solved. When the fitness function of the DBO algorithm finds its optimal solution, the corresponding optimal solution is the optimal hyperparameter combination of the XGBoost algorithm. Using the DBO algorithm for optimization can greatly save the time spent manually tuning and finding the optimal hyperparameters, and the optimal hyperparameter combination can further improve the accuracy and generalization ability of the XGBoost algorithm for battery temperature estimation. This embodiment uses the RMSE of temperature estimation as its standard. The smaller the RMSE value, the more accurate the XGBoost algorithm's estimation result of the battery temperature. Based on the above analysis, this embodiment sets the fitness function of the DBO optimization algorithm as follows:

[0094]

[0095] After the above iterative optimization, the global optimal solution of the algorithm is now found. This refers to the optimal hyperparameters of the XGBoost algorithm. By searching for the optimal hyperparameter combination of the XGBoost algorithm using the DBO algorithm, the accuracy and generalization ability of the proposed battery temperature estimation method are significantly improved. The flowchart of the method is shown below. Figure 1 The left half is shown. The specific steps are as follows:

[0096] 1) Perform EIS testing on each cell in the lithium-ion battery pack, conduct correlation analysis on the obtained EIS data, extract relevant features based on the correlation analysis results, and establish a sample dataset.

[0097] 2) Divide the sample dataset into training and test sets according to the proportion. The training set data is further divided into training and validation sets according to the k-fold cross-validation method.

[0098] 3) Input the training and validation sets into the XGBoost algorithm, use the DBO algorithm to search for the optimal hyperparameter combination of the XGBoost algorithm, and input the optimal hyperparameters into the XGBoost algorithm. Finally, a lithium-ion battery temperature estimation method based on the DBO-XGBoost algorithm is obtained.

[0099] 4) Input the test set data into the DBO-XGBoost algorithm that has been optimized and trained in step 3), and use the proposed method to estimate the battery temperature.

[0100] Based on the correlation analysis results, feature extraction was performed on the collected battery EIS data. The dataset was divided into training and testing sets in an 8:2 ratio, and 5-fold cross-validation was used for both training and validation on the training set. The training and validation sets were then input into the DBO optimization algorithm, with a population size of 10 and a maximum number of iterations of 100. Optimization was performed within the set parameter range, and the convergence curve of the fitness function is shown below. Figure 20 As shown.

[0101] Depend on Figure 20 It can be seen that the fitness function value of the DBO optimization algorithm used in this invention, that is, the RMSE value of the XGBoost algorithm for battery temperature estimation using the training set, converged to 1.108 in less than 20 iterations, indicating that the algorithm has a strong optimization ability in the hyperparameter optimization of the XGBoost algorithm.

[0102] The optimal hyperparameter combination obtained by the DBO algorithm is input into the XGBoost algorithm. Data from the training set is then input into the DBO-optimized XGBoost algorithm for training. Test set data is then used as input, and finally, the battery temperature estimation result is output. The RMSE, MAE, MBE, and R² obtained by this method are 1.43, 1.19, -0.14, and 0.97, respectively.

[0103] In this embodiment, during subsequent use, by measuring the electrochemical impedance spectroscopy at the unknown temperature and inputting it into the optimized and trained DBO-XGBoost model, the temperature estimate of the battery under test can be obtained. .

[0104] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for estimating the temperature of a lithium-ion battery based on electrochemical impedance spectroscopy, characterized in that, Includes the following steps: Step 1: Measure the EIS data of the same type of lithium-ion battery at different SOCs, different charge-discharge cycles, and different temperatures; Step 2: Perform feature analysis on the EIS data obtained in Step 1, and select one or more feature frequencies that meet the requirements of correlation with SOC, charge-discharge cycle number and temperature through statistical and correlation analysis. Step 3: Extract the electrochemical impedance spectroscopy data obtained in Step 1 according to the characteristic frequencies determined in Step 2, and select characteristic variables based on the extraction results and correlation analysis; Step 4: Preprocess the dataset obtained in Step 3, and use dung beetle optimization to globally optimize the hyperparameters of the XGBoost model and train the model. Finally, input the EIS features of the test set into the model to output the battery temperature estimate.

2. The lithium-ion battery temperature estimation method based on electrochemical impedance spectroscopy according to claim 1, characterized in that, Step 1 is as follows: Step 1-1: Take batteries of the same model that have been cycled 0, 100, 200, 300, and 400 times, charge the batteries to 100% SOC at 25°C using the CC-CV method, and let them stand for 4 hours. Steps 1-2: Discharge the battery with a current of 0.5C for 12 minutes to discharge the battery to 90% SOC. Then, place the battery in a high and low temperature test chamber set to 25℃ and let it stand for 8 hours. Steps 1-3: Perform EIS testing on the battery using an electrochemical workstation; Steps 1-4: Increase the temperature of the high and low temperature test chamber by 5°C and let it stand for another 8 hours. Steps 1-5, repeat steps 1-3 to 1-4, until the temperature reaches 50℃; Steps 1-6: Reset the high and low temperature test chamber to 25℃ and let it stand for 8 hours; Steps 1-7: Discharge the battery with a current of 0.5C for 24 minutes, then let it stand for 8 hours; Steps 1-8, repeat steps 1-3 to 1-7, until the loop reaches 10 times.

3. The lithium-ion battery temperature estimation method based on electrochemical impedance spectroscopy according to claim 1, characterized in that, Step 2 is as follows: The PCC method is used to quantitatively analyze the correlation between data from various characteristic points and battery SOC, temperature, and charge-discharge cycle count; the PCC formula is defined as: ； in, Let (X, Y) be the covariance between variables. The product of the standard deviations of variables (X,Y) is used. The PCC value is between [-1,1]. When its value is close to ±1, there is a strong positive and negative correlation between variables X and Y. The smaller the absolute value, the weaker the correlation between the two variables. The correlation between impedance parameters and SOC, charge-discharge cycle number and temperature at each frequency point is quantified by Pearson correlation coefficient. The characteristic frequencies of impedance imaginary part and phase with SOC |PCC|≤0.65, with charge-discharge cycle number |PCC|≤0.65, and with temperature |PCC|≥0.94 are selected.

4. The lithium-ion battery temperature estimation method based on electrochemical impedance spectroscopy according to claim 1, characterized in that, Step 3 specifically involves: The electrochemical impedance spectroscopy data obtained in step 1 are extracted according to the characteristic frequencies determined in step 2. Combined with correlation analysis, the imaginary part of the battery impedance and the phase value measured at the characteristic frequencies are selected as characteristic variables.

5. The method for estimating the temperature of a lithium-ion battery based on electrochemical impedance spectroscopy according to claim 1, characterized in that, Step 4 specifically involves: The dataset obtained in step 3 is divided into a training set and a test set according to a predetermined ratio. The training set is divided into a training subset and a validation subset in each round through k-fold cross-validation for hyperparameter search and model selection. The training and validation subsets are input into the XGBoost regressor. Based on this, the DBO algorithm is introduced to adaptively optimize the key hyperparameters of XGBoost globally. The optimization objective is the temperature estimation error on the validation set. The key hyperparameters include learning rate, maximum tree depth, subsample, colsample bytree, and number of base learners n_estimators. The optimal hyperparameters obtained through DBO search were used to train the final XGBoost model, forming the DBO-XGBoost temperature estimation model. The reserved test set is input into the optimized and trained DBO-XGBoost model to evaluate and output the temperature estimation results and performance indicators of the lithium-ion battery.

6. A lithium-ion battery temperature estimation system based on electrochemical impedance spectroscopy, characterized in that, include: The first module is used to perform EIS testing on lithium-ion batteries of the same model at different SOCs, different charge-discharge cycles, and different temperatures to obtain the corresponding electrochemical impedance spectroscopy data. The second module, based on the battery electrochemical impedance spectroscopy, selects one or more characteristic frequencies that meet the requirements for correlation with SOC, charge-discharge cycle number and temperature through statistical and correlation analysis. The third module, in conjunction with the characteristic frequencies determined in the second module, extracts the EIS characteristic parameters at the corresponding frequency points and, in conjunction with correlation analysis, determines the characteristic variables suitable for temperature estimation. The fourth module uses DBO to adaptively optimize the hyperparameters of the XGBoost model globally and completes model training based on the training set data; the EIS feature variables of the test set are input into the optimized and trained model to output the temperature estimation results of the lithium-ion battery.

7. The lithium-ion battery temperature estimation system based on electrochemical impedance spectroscopy according to claim 6, characterized in that, The first module performs EIS testing on lithium-ion batteries of the same model at different SOCs, different charge-discharge cycle numbers, and different temperatures to obtain corresponding electrochemical impedance spectroscopy data. Specifically: 1) Take batteries of the same model that have been cycled 0, 100, 200, 300, and 400 times, charge the batteries to 100% SOC at 25℃ using the CC-CV method, and let them stand for 4 hours. 2) Discharge the battery with a current of 0.5C for 12 minutes until the battery is discharged to 90% SOC. Then place the battery in a high and low temperature test chamber set to 25℃ and let it stand for 8 hours. 3) Perform EIS testing on the battery using an electrochemical workstation; 4) Increase the temperature of the high and low temperature test chamber by 5°C and let it stand for another 8 hours; 5) Repeat steps 3) to 4) until the temperature reaches 50℃; 6) Reset the high and low temperature test chamber to 25℃ and let it stand for 8 hours; 7) Discharge the battery at a current of 0.5C for 24 minutes, then let it stand for 8 hours; 8) Repeat steps 3) to 7) until the loop reaches 10 times.

8. The lithium-ion battery temperature estimation system based on electrochemical impedance spectroscopy according to claim 6, characterized in that, The second module, based on the battery electrochemical impedance spectroscopy, selects one or more characteristic frequencies that meet the requirements for correlation with SOC, charge-discharge cycle count, and temperature through statistical and correlation analysis. Specifically: The PCC method is used to quantitatively analyze the correlation between data from various characteristic points and battery SOC, temperature, and charge-discharge cycle count; the PCC formula is defined as: ； in, Let (X, Y) be the covariance between variables. The PCC is the product of the standard deviations of variables (X,Y). The value of PCC is between [-1,1]. When its value is close to ±1, there is a strong positive and negative correlation between variables X and Y. The smaller the absolute value, the weaker the correlation between the two variables. The correlation between impedance parameters and SOC, charge-discharge cycle number and temperature at each frequency point is quantified by PCC. The characteristic frequencies of impedance imaginary part and phase with SOC |PCC|≤0.65, with charge-discharge cycle number |PCC|≤0.65, and with temperature |PCC|≥0.94 are selected.

9. The lithium-ion battery temperature estimation system based on electrochemical impedance spectroscopy according to claim 6, characterized in that, The third module is specifically as follows: The electrochemical impedance spectroscopy data obtained in the first module are extracted according to the characteristic frequencies determined in the second module. Combined with correlation analysis, the imaginary part of the battery impedance and the phase value measured at the characteristic frequencies are selected as characteristic variables.

10. The lithium-ion battery temperature estimation system based on electrochemical impedance spectroscopy according to claim 6, characterized in that, The fourth module is specifically as follows: The dataset obtained in the third module is divided into a training set and a test set according to a predetermined ratio. The training set is divided into a training subset and a validation subset in each round through k-fold cross-validation for hyperparameter search and model selection. The training and validation subsets are input into the XGBoost regressor. Based on this, the DBO algorithm is introduced to adaptively optimize the key hyperparameters of XGBoost globally. The optimization objective is the temperature estimation error on the validation set. The key hyperparameters include learning rate, maximum tree depth, subsample, colsample bytree, and number of base learners n_estimators. The optimal hyperparameters obtained through DBO search were used to train the final XGBoost model, forming the DBO-XGBoost temperature estimation model. The reserved test set is input into the optimized and trained DBO-XGBoost model to evaluate and output the temperature estimation results and performance indicators of the lithium-ion battery.

Images