Lithium battery health state evaluation method and system based on AI algorithm

Abstract

The invention relates to the technical field of lithium battery health assessment, in particular to a lithium battery health state assessment method and system based on an AI algorithm, and the method comprises the steps: obtaining battery data and aging mechanism parameters based on a pulse curve and an input current; calculating the available capacity of the battery according to the battery aging mechanism parameters; daily data of the battery are collected, and an SVM model is trained in cooperation with aging mechanism parameters; calculating the number of remaining available days and the number of recommended replacement days of the battery based on an SVM model; therefore, the available capacity of the battery is calculated according to the battery aging mechanism parameters, a user can visually check the health state of the battery, meanwhile, the remaining available days and the recommended replacement days are predicted, the remaining life and the replacement date are predicted according to the habits and the available capacity of the user, the prediction accuracy is ensured, and the user experience is improved. And the situation that the lithium battery is not replaced for a long time, the health state is too poor, and the use experience is affected is avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of lithium battery health assessment, and particularly to a method and system for assessing the health state of a lithium battery based on an AI algorithm. Background Art

[0002] At present, with the continuous increase in the demand for electrochemical energy storage in new power systems, in order to meet the voltage and power requirements of energy storage systems, lithium battery energy storage systems need to be composed of hundreds or thousands of battery monomers connected in series and parallel for use. For large-scale battery systems, the system may be composed of different types, different magnitudes, and different echelon stacks, and the inconsistency of system parameters and performance will be more obvious, resulting in a more serious difference in the decay rate of the core life. The decay of lithium battery life is a dynamic electrochemical process, and the decay mainly comes from the loss of active lithium, the loss of active materials, and polarization loss.

[0003] In the aforementioned prior art, in order to improve the safety of lithium batteries, the safety state of lithium batteries is usually detected. However, in terms of the health state of lithium batteries, there is a lack of functions for predicting the remaining available life of lithium batteries and the specific date when replacement is required. Therefore, it will lead to unclear how to maintain or replace and maintain lithium batteries for staff, affecting the use experience of lithium batteries. Summary of the Invention

[0004] The purpose of the present invention is to provide a method and system for assessing the health state of a lithium battery based on an AI algorithm, which solves the problem that in the prior art, in order to improve the safety of lithium batteries, the safety state of lithium batteries is usually detected. However, in terms of the health state of lithium batteries, there is a lack of functions for predicting the remaining available life of lithium batteries and the specific date when replacement is required. Therefore, it will lead to unclear how to maintain or replace and maintain lithium batteries for staff, affecting the use experience of lithium batteries.

[0005] To achieve the above purpose, the present invention provides a method for assessing the health state of a lithium battery based on an AI algorithm, which obtains battery data and aging mechanism parameters based on a pulse curve and input current; Calculates the available capacity of the battery according to the battery aging mechanism parameters; Collects the daily data of the battery, and trains an SVM model in combination with the aging mechanism parameters; Calculates the remaining available days and recommended replacement days of the battery based on the SVM model.

[0006] Among them, in the step of calculating the available capacity of the battery according to the battery aging mechanism parameters: Performs sensitivity analysis on the aging mechanism parameters to obtain sensitivity parameters; Identifies the identification parameters of the four aging mechanism parameters based on the sensitivity parameters; Combined with four identification parameters, the available capacity of the battery is calculated in real time.

[0007] Among them, the steps of collecting the daily data of the battery and training the SVM model in combination with the aging mechanism parameters specifically include: Collect daily usage habit data; Perform data preprocessing on the collected daily data, aging mechanism parameters and identification data; Randomly select 70% of the data to construct a training set, and the remaining 30% of the data to construct a test set; Use SVM to map the data set to a high-dimensional space through a radial basis function kernel to find the optimal hyperplane; Set the parameters of SVM based on the hyperplane; Use the training set data, the selected function kernel and the set parameters to train the SVM model.

[0008] Among them, the steps of calculating the remaining available days and recommended replacement days of the battery based on the SVM model specifically include: Perform real-time detection on the battery to obtain battery capacity data; Perform preprocessing on the battery capacity data to make its data format the same as that of the training set; Input the battery capacity data into the SVM model; Based on the representation of the input battery capacity data in the high-dimensional space and the position of the optimal hyperplane, the SVM model calculates the predicted value of the daily consumption capacity of the battery; Divide the available capacity of the battery by the predicted value of the daily consumption capacity of the battery to obtain the remaining available prediction range and the recommended replacement prediction range; Repeat the calculation of the four prediction ranges of the remaining available prediction range and the recommended replacement prediction range, and take the average value to obtain the remaining available days and the recommended replacement days respectively.

[0009] Among them, the steps of performing sensitivity analysis on the aging mechanism parameters to obtain sensitivity parameters specifically include: Select the data when the battery is in good health as the reference point, and record four battery mechanism parameters and the corresponding battery performance indicators; Select one battery mechanism parameter, and use a ±5% value range as the change amount, and perform the change value of the aging mechanism parameter based on the change amount; Keep the other battery mechanism parameters unchanged, change the value of each battery mechanism parameter one by one, and observe the change value of the battery performance indicator; Record the change value of the battery performance indicator when each mechanism parameter changes to obtain the sensitivity parameter.

[0010] Among them, the steps of preprocessing the collected daily data, aging mechanism parameters, and identification data specifically include: Using statistical methods to judge outliers, duplicate values, and missing values, and deleting outliers and duplicate values from the data; Using interpolation to fill in the data points of missing values; Converting the data to a standard normal distribution by subtracting the mean and dividing by the standard deviation.

[0011] The present invention also provides a lithium battery state of health assessment system based on an AI algorithm, which adopts the above-mentioned lithium battery state of health assessment method based on an AI algorithm, and includes a current input module, an available capacity calculation module, a model construction module, and a days prediction module; The current input module is used to obtain battery data and aging mechanism parameters based on the pulse curve and input current; The available capacity calculation module is used to calculate the available capacity of the battery according to the battery aging mechanism parameters; The model construction module is used to collect the daily battery data and cooperate with the aging mechanism parameters to train an SVM model; The days prediction module is used to calculate the remaining available days and recommended replacement days of the battery based on the SVM model.

[0012] A lithium battery state of health assessment method and system based on an AI algorithm of the present invention obtains battery data and aging mechanism parameters based on the pulse curve and input current; calculates the available capacity of the battery according to the battery aging mechanism parameters; collects the daily battery data and cooperates with the aging mechanism parameters to train an SVM model; calculates the remaining available days and recommended replacement days of the battery based on the SVM model; Thus, relying on the battery aging mechanism parameters to calculate the available capacity of the battery, the user can intuitively view the battery state of health, and at the same time make predictions on the remaining available days and recommended replacement days, predict the remaining life and replacement date according to the user's habits and available capacity, ensure the accuracy of the prediction, so that the staff can perform regular maintenance or replacement, and avoid the poor state of health of the lithium battery due to long-term non-replacement affecting the use experience. Description of the Drawings

[0013] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required to be used in the description of the embodiments or the prior art.

[0014] Figure 1 It is a multi-stage multi-pulse charging current curve diagram of the present invention.

[0015] Figure 2 It is an analysis diagram of the sensitivity of battery electrochemical aging parameters of the present invention.

[0016] Figure 3 It is a flowchart of the steps of the lithium battery state of health assessment method based on AI algorithm of the present invention.

[0017] Figure 4 It is a flowchart of the steps of calculating the available capacity of the battery according to the battery aging mechanism parameters of the present invention.

[0018] Figure 5 It is a flowchart of the steps of collecting the daily data of the battery, cooperating with the aging mechanism parameters, and training an SVM model of the present invention.

[0019] Figure 6 It is a flowchart of the steps of calculating the remaining available days and recommended replacement days of the battery based on the SVM model of the present invention.

[0020] Figure 7 It is a flowchart of the steps of performing sensitivity analysis on the aging mechanism parameters to obtain sensitivity parameters of the present invention.

[0021] Figure 8 It is a flowchart of the steps of performing data preprocessing on the collected daily data, aging mechanism parameters, and identification data of the present invention. Detailed implementation manners

[0022] The embodiments of the present invention are described in detail below. The examples of the embodiments are shown in the accompanying drawings. The embodiments described below by referring to the accompanying drawings are exemplary and are intended to explain the present invention and should not be construed as limiting the present invention.

[0023] Please refer to Figures 1 to 8 , the present invention provides a lithium battery state of health assessment method based on AI algorithm, including the following steps: S101: Obtain battery data and aging mechanism parameters based on the pulse curve and input current; The pulse curve refers to the multi-stage multi-pulse charging current curve diagram of the accompanying drawings of the present application. The input current length is 1000 s, and the maximum charging current is 1.6 C; the battery data includes voltage and temperature, and the aging mechanism parameters include the maximum lithium ion concentration of the battery negative electrode, the maximum lithium ion concentration of the battery positive electrode, the percentage of the active material of the battery negative electrode, and the percentage of the active material of the battery positive electrode.

[0024] S102: Calculate the available capacity of the battery according to the battery aging mechanism parameters; Specifically include: S1021: Perform sensitivity analysis on the aging mechanism parameters to obtain sensitivity parameters; Specifically include: S10211: Select the data when the battery is in good health state as the reference point, and record four battery mechanism parameters and the corresponding battery performance indicators; By recording the mechanism parameters and performance indicators of the battery in good health, a standard reference range can be established. This helps in the subsequent monitoring and evaluation of the battery state, ensuring that the battery always remains in the optimal working state; these benchmark data can be used to evaluate the current performance state of the battery. By comparing with the benchmark data, a decline or abnormality in battery performance can be detected in a timely manner, and corresponding maintenance or replacement measures can be taken.

[0025] S10212: Select a battery mechanism parameter, and use a ±5% value range as the variation amount to perform the variation value of the aging mechanism parameter based on the variation amount; Assume that the initial capacity of the battery in good health is C0. Using a ±5% value range as the variation amount, that is, the battery capacity can vary between 95% of C0 (C0 - 5%C0) and 105% of C0 (C0 + 5%C0).

[0026] S10213: Keep other battery mechanism parameters unchanged, change the value of each battery mechanism parameter one by one, and observe the variation value of the battery performance indicator; When changing the value of the battery mechanism parameter one by one, it will be found that the battery performance indicators (such as energy density, power density, cycle life, etc.) change significantly. These changes reflect the direct impact of parameter adjustment on battery performance; based on the experimental results of the control variable method, the battery can be optimized in design. The optimized battery usually has a significant improvement in performance indicators.

[0027] S10214: Record the variation value of the battery performance indicator when each mechanism parameter changes to obtain the sensitivity parameter.

[0028] Record the variation values of the battery performance indicators in the above steps in sequence, and finally obtain the sensitivity of each parameter, referring to the battery electrochemical aging parameter sensitivity analysis diagram of this application.

[0029] S1022: Perform identification based on the sensitivity parameter to obtain the identification parameters of the four aging mechanism parameters; Based on the electrochemical principle equation of the battery, combined with known conditions such as voltage, current, and temperature, calculate the sensitive parameters in sequence. The parameter calculation objective function is as follows, that is, by finding a set of optimal parameter solutions to minimize the error between the measured terminal voltage and the model terminal voltage: where, θ * — The optimal parameter set, by minimizing the experimental voltage and the simulated voltage to solve, the goal is to find a set of parameters to minimize the sum of the squares of the errors between the experimental data and the model prediction values; — The maximum lithium ion concentration of the negative electrode material; — The effective volume fraction of the negative electrode, representing the proportion of the active material in the negative electrode.

[0030] — The maximum lithium ion concentration of the positive electrode material; — The effective volume fraction of the positive electrode, representing the proportion of the active material in the positive electrode.

[0031] S1023: Combine the four identification parameters to calculate the available battery capacity in real time.

[0032] The battery capacity calculation formula is as follows: The formula for calculating the negative electrode capacity: Among them, Q neg — is the capacity of the negative electrode, Ah; A — The plate area; F — The Faraday constant; L neg — The thickness of the negative electrode plate; — The change value of the electrochemical equivalent of the negative electrode.

[0033] Calculate the positive electrode capacity: Among them, Q pos — is the capacity of the positive electrode, Ah; A — The plate area; F — The Faraday constant; L pos — The thickness of the positive electrode plate; — The change value of the electrochemical equivalent of the positive electrode.

[0034] The maximum actually available battery capacity is the smaller value of the two: ; Combined with the four electrochemistry calculation parameters ( ), the available battery capacity can be calculated in real time.

[0035] S103: Collect the daily data of the battery, and train an SVM (Support Vector Machine) model in combination with the aging mechanism parameters; Specifically include: S1031: Collect the daily usage habit data; Daily habit data includes battery usage records, usage duration, usage behavior, usage intensity, ambient temperature and humidity, and discharge capacity. Collecting a large amount of habit data can effectively reflect the user's daily usage habits and behaviors, and thus more accurately predict the remaining number of days of battery life.

[0036] S1032: Perform data preprocessing on the collected daily data, aging mechanism parameters, and identification data; Specifically including: S10321: Use statistical methods to judge outliers, duplicate values, and missing values, and delete outliers and duplicate values from the data; Outliers refer to data points in a dataset that deviate significantly from other observed values. These values may be caused by measurement errors, data entry errors, natural variations, or extreme events. Statistical methods identify outliers by calculating the distribution characteristics (mean) of the data, setting thresholds, or using graphical tools (such as box plots); identify records with exactly the same or partially the same content by comparing the records in the dataset; improve the accuracy and consistency of the data.

[0037] S10322: Use interpolation to fill in the missing data points; Construct a straight line through two known data points to estimate the missing data; calculation formula: y = y1 + (y2 - y1) / (x2 - x1) × (x - x1), where y1 and y2 are the values of the known data points, x1 and x2 are the independent variable values corresponding to the known data points, x is the independent variable value corresponding to the missing data, and y is the missing data value estimated by linear interpolation; can effectively fill in the missing part of the data, make the dataset more complete, and thus meet the needs of data analysis, modeling, and visualization; can reduce the blanks and noise in the data, and improve the accuracy and reliability of the data.

[0038] S10323: Convert the data to a standard normal distribution by subtracting the mean and dividing by the standard deviation.

[0039] Realize the standardization of the data, which helps to eliminate the influence of the data dimension, so that different features can be compared on the same scale; the standardized data has a unified scale and distribution characteristics, which helps to improve the accuracy of data analysis; the data of the standard normal distribution has the characteristics of a mean of 0 and a standard deviation of 1, which makes the comparison and analysis between data points more intuitive and convenient; The principle of converting the data to a standard normal distribution is based on linear transformation, that is, subtracting the mean of each value of the original data and dividing by the standard deviation. The core idea of this method is to adjust the position and scale of the data so that the converted data follows a standard normal distribution with a mean of 0 and a standard deviation of 1. The conversion formula is: Z = (X - μ) / σ Where Z is the random variable of the transformed standard normal distribution, X is the random variable of the original data, μ is the mean of the original data, and σ is the standard deviation of the original data.

[0040] S1033: Randomly select 70% of the data to construct a training set, and the remaining 30% of the data to construct a test set; By dividing into two data sets, it can ensure that the algorithm is trained and tested on independent data sets, so as to evaluate its generalization performance; random selection guarantees the independence between the training set and the test set, avoiding the risk of data leakage. The training set is used to train the model, and the test set is used to test the model; by evaluating the performance of the algorithm on an independent test set, more accurate and reliable evaluation results can be obtained. This helps algorithm developers understand the advantages and disadvantages of the algorithm and make targeted optimizations and improvements.

[0041] S1034: Use SVM to map the data set to a high-dimensional space through the radial basis function kernel to find the optimal hyperplane; The radial basis function kernel (RBF kernel) is a commonly used kernel function in SVM. It can map the samples in the input space to a high-dimensional feature space, thus realizing nonlinear classification and regression; for each training sample, calculate the distance between this sample and other samples, and calculate the kernel value according to the definition of the kernel function. These kernel values form a kernel matrix for subsequent classification or regression tasks; in the feature space, the SVM algorithm finds the optimal hyperplane by solving a quadratic programming problem. This hyperplane can maximize the margin between different classes, thus improving the classification accuracy and generalization ability; once the optimal hyperplane is found, it can be used to classify new input samples. For a given input sample, calculate its distance to the hyperplane and judge the class to which the sample belongs according to the sign of the distance.

[0042] S1035: Set the parameters of SVM based on the hyperplane; The parameter setting of SVM directly affects the position and shape of the hyperplane. For example, the penalty parameter C and the parameters of the kernel function (such as the gamma parameter of the RBF kernel) will affect the decision boundary of the hyperplane; the penalty parameter C is used to control the penalty degree for misclassified samples. The larger the value of C, the greater the penalty for misclassification, which may cause the hyperplane to fit the training data more closely, but may reduce the generalization ability; the smaller the value of C, the smaller the penalty for misclassification, allowing a certain error, which may improve the generalization ability; by carefully setting the parameters of SVM, an optimal hyperplane can be found to maximize the classification accuracy.

[0043] S1036: Use the training set data, the selected function kernel, and the set parameters to train an SVM model.

[0044] Train the SVM model for subsequent prediction of the remaining available days.

[0045] S104: Calculate the remaining available days and the recommended replacement days of the battery based on the SVM model.

[0046] Specifically, it includes: S1041: Conduct real-time detection on the battery to obtain battery capacity data; By collecting a large amount of data from the battery, the current usage situation of the battery can be more comprehensively reflected, improving the accuracy of subsequent predicted days.

[0047] S1042: Preprocess the battery capacity data to make its data format the same as that of the training set; By using the same processing method for the battery capacity data as the above-mentioned daily habit data, outliers and duplicate values are removed, missing values are filled, errors and uncertainties in the model training process are reduced, and the accuracy and generalization ability of the model are improved; the processed data has a consistent format and high quality, which helps to speed up the model training. At the same time, through techniques such as feature extraction and data augmentation, the time and resources required for model training can be further reduced, which helps to improve the performance of the model.

[0048] S1043: Input the battery capacity data into the SVM model; Input the preprocessed data as above and perform subsequent model calculations to obtain predicted values.

[0049] S1044: The SVM model calculates the predicted value of the daily battery consumption capacity based on the representation of the input battery capacity data in the high-dimensional space and the position of the optimal hyperplane; After being trained, the SVM model has mastered the rules and characteristics of the battery. At this time, the latest battery capacity data is input, and then based on the representation in the high-dimensional space and the position of the optimal hyperplane, the predicted value of the daily battery consumption capacity is calculated.

[0050] S1045: Divide the available capacity of the battery by the predicted value of the daily battery consumption capacity to obtain the remaining available prediction range and the recommended replacement prediction range; Divide the available capacity of the battery obtained from the above formula by the predicted value of the daily battery consumption capacity predicted by the above model to obtain the remaining available prediction range, that is, the remaining available days. Arrange the remaining available days in sequence to obtain the recommended replacement prediction range, that is, recommend a certain day in the future for replacement.

[0051] S1046: Repeatedly calculate the four prediction ranges of the remaining available prediction range and the recommended replacement prediction range, and take the average value to obtain the remaining available days and the recommended replacement days respectively.

[0052] Through the above range calculation method, repeat it four times, and respectively take the average values within the four ranges of the remaining available prediction range and the recommended replacement prediction range, so that the obtained number of days is more accurate and large deviations are avoided; the obtained remaining available days represent how many days the battery can still be used according to the current usage habits of the user; the obtained recommended replacement days represent on which day in the future the battery needs to be replaced according to the current usage habits of the user, otherwise the battery health status will decline and the performance will be greatly reduced.

[0053] The present invention also provides a lithium battery health status evaluation system based on an AI algorithm, which adopts the lithium battery health status evaluation method based on the AI algorithm, and includes a current input module, an available capacity calculation module, a model construction module, and a number of days prediction module; the current input module is used to obtain battery data and aging mechanism parameters based on the pulse curve and the input current; the available capacity calculation module is used to calculate the available capacity of the battery according to the battery aging mechanism parameters; the model construction module is used to collect the daily data of the battery and train an SVM model in cooperation with the aging mechanism parameters; the number of days prediction module is used to calculate the remaining available days and the recommended replacement days of the battery based on the SVM model.

[0054] Among them, the current input module is used to obtain battery data and aging mechanism parameters based on the pulse curve and the input current; the pulse curve refers to the multi-stage multi-pulse charging current curve diagram in the attached drawings of this application, with an input current length of 1000 s and a maximum charging current of 1.6 C; the battery data includes voltage and temperature, and the aging mechanism parameters include the maximum lithium ion concentration at the negative electrode of the battery, the maximum lithium ion concentration at the positive electrode of the battery, the percentage of active material at the negative electrode of the battery, and the percentage of active material at the positive electrode of the battery. The available capacity calculation module is used to calculate the available capacity of the battery according to the battery aging mechanism parameters; perform sensitivity analysis on the aging mechanism parameters to obtain sensitivity parameters, select the data when the battery is in good health as the reference point, record the four battery mechanism parameters and the corresponding battery performance indicators. By recording the mechanism parameters and performance indicators of the battery in good health, a standard reference range can be established. This helps the subsequent monitoring and evaluation of the battery state to ensure that the battery always remains in the best working state; these reference data can be used to evaluate the current performance state of the battery, select a battery mechanism parameter, and use a ±5% value range as the change amount. Based on the change amount, the change value of the aging mechanism parameter is carried out, while keeping other battery mechanism parameters unchanged. The value of each battery mechanism parameter is changed one by one, and the change value of the battery performance indicator is observed; when the value of the battery mechanism parameter is changed one by one, it will be found that the battery performance indicators (such as energy density, power density, cycle life, etc.) change significantly. These changes reflect the direct impact of parameter adjustment on battery performance. Record the change value of the battery performance indicator when each mechanism parameter changes to obtain the sensitivity parameter; based on the sensitivity parameter, perform identification to obtain the identification parameters of the four aging mechanism parameters, and finally calculate the available capacity of the battery in real time. The model construction module is used to collect the daily data of the battery, cooperate with the aging mechanism parameters, and train an SVM model; collect the daily usage habit data; perform data preprocessing on the collected daily data, aging mechanism parameters, and identification data, use statistical methods to judge outliers, duplicate values, and missing values, and delete the outliers and duplicate values from the data, and use the interpolation method to fill in the data points of the missing values. The data is transformed into a standard normal distribution by subtracting the mean and dividing by the standard deviation; the standardization of the data is realized, which helps to eliminate the influence of the dimension of the data, so that different features can be compared on the same scale; the standardized data has a unified scale and distribution characteristics, which helps to improve the accuracy of data analysis. Randomly select 70% of the data to construct a training set, and the remaining 30% of the data to construct a test set. Use SVM to map the data set to a high-dimensional space through the radial basis function kernel to find the optimal hyperplane, set the parameters of SVM based on the hyperplane, and use the training set data, the selected function kernel, and the set parameters to train an SVM model.The number of days prediction module is used to calculate the remaining available days and the recommended replacement days of the battery based on the SVM model; after being trained, the SVM model has mastered the laws and characteristics of the battery. At this time, the latest battery capacity data is input, and then the representation in the high-dimensional space and the position of the optimal hyperplane are used to calculate the predicted value of the daily battery consumption capacity; the available capacity of the battery is divided by the predicted value of the daily battery consumption capacity to obtain the remaining available prediction range and the recommended replacement prediction range; the four prediction ranges of the remaining available prediction range and the recommended replacement prediction range are calculated repeatedly, and the average value is taken to obtain the remaining available days and the recommended replacement days respectively.

[0055] The above-disclosed are only one or more preferred embodiments of the present application, and the scope of rights of the present application cannot be limited thereby. Those of ordinary skill in the art can understand all or part of the processes of implementing the above embodiments, and the equivalent changes made according to the claims of the present application still fall within the scope covered by the present application.

Claims

1. A lithium battery health status assessment method based on AI algorithm, characterized in that: The steps include: Based on the pulse curve and input current, the battery data and aging mechanism parameters are obtained; Calculate the available capacity of the battery based on the battery aging mechanism parameters; Collect daily battery data and train the SVM model based on the aging mechanism parameters; The remaining usable days of the battery and the recommended replacement days are calculated based on the SVM model.

2. The lithium battery health status assessment method based on AI algorithm according to claim 1, characterized in that: The step of calculating the available capacity of the battery according to the battery aging mechanism parameters specifically includes: Conduct sensitivity analysis on aging mechanism parameters to obtain sensitivity parameters; Based on the sensitivity parameters, identification parameters of four aging mechanism parameters are obtained; Combining the four identification parameters, the available battery capacity is calculated in real time.

3. The lithium battery health status assessment method based on AI algorithm as claimed in claim 2, characterized in that: The steps of collecting daily battery data and training the SVM model in combination with the aging mechanism parameters specifically include: Collect data on daily usage habits; Perform data preprocessing on the collected daily data, aging mechanism parameters and identification data; 70% of the data is randomly selected to construct the training set, and the remaining 30% of the data is used to construct the test set; Use SVM to map the data set into a high-dimensional space through a radial basis function kernel to find the optimal hyperplane; Set the parameters of SVM based on the hyperplane; The SVM model is trained using the training set data, the selected function kernel and the set parameters.

4. The lithium battery health status assessment method based on AI algorithm as claimed in claim 3, characterized in that: The step of calculating the remaining usable days of the battery and the recommended replacement days based on the SVM model specifically includes: Conduct real-time battery testing to obtain battery capacity data; Preprocess the battery capacity data to make it the same as the data format of the training set; Inputting the battery capacity data into the SVM model; The SVM model calculates the predicted value of the daily battery consumption capacity according to the representation of the input battery capacity data in the high-dimensional space and the position of the optimal hyperplane; Divide the battery available capacity by the predicted value of the battery daily consumption capacity to obtain a remaining available prediction range and a recommended replacement prediction range; Repeat the calculation of the four prediction ranges of the remaining available prediction range and the recommended replacement prediction range, and take the average value to obtain the remaining available days and the recommended replacement days respectively.

5. The lithium battery health status assessment method based on AI algorithm as claimed in claim 4, characterized in that: The step of performing sensitivity analysis on the aging mechanism parameters to obtain the sensitivity parameters specifically includes: Select the data when the battery is in a good health state as the benchmark point, and record the four battery mechanism parameters and the corresponding battery performance indicators; Select a battery mechanism parameter, and use a value range of ±5% as a variation, and change the value of the aging mechanism parameter based on the variation; Keep other battery mechanism parameters unchanged, change the value of each battery mechanism parameter one by one, and observe the change value of battery performance index; The change value of the battery performance index when each mechanism parameter changes is recorded to obtain the sensitivity parameter.

6. The lithium battery health status assessment method based on AI algorithm as claimed in claim 5, characterized in that: The step of preprocessing the collected daily data, aging mechanism parameters and identification data specifically includes: Statistical methods were used to determine outliers, duplicate values, and missing values, and outliers and duplicate values ​​were deleted from the data; Interpolation was used to fill in missing data points; The data were transformed to a standard normal distribution by subtracting the mean and dividing by the standard deviation.

7. A lithium battery health status assessment system based on an AI algorithm, using the lithium battery health status assessment method based on an AI algorithm as claimed in claim 6, characterized in that: It includes current input module, available capacity calculation module, model building module and days prediction module; The current input module is used to obtain battery data and aging mechanism parameters based on the pulse curve and input current; The available capacity calculation module is used to calculate the available capacity of the battery according to the battery aging mechanism parameters; The model building module is used to collect daily battery data and train the SVM model in combination with the aging mechanism parameters; The days prediction module is used to calculate the remaining usable days of the battery and the recommended replacement days based on the SVM model.

Images