Lithium battery SOC estimation algorithm, system, equipment and medium based on CSHHO-LSSVM

Abstract

The present invention discloses a lithium battery SOC estimation algorithm, system, device, and medium based on CSHHO-LSSVM, which relates to the field of battery state of charge estimation. The algorithm includes collecting SOC data of lithium phosphate batteries to form a sample set, normalizing it, and setting each initial parameter; initializing a population based on seven chaotic maps, calculating individual fitness values, and calculating escape energy; updating the population in the algorithm's exploration phase according to a position update formula, and introducing a variable spiral position update strategy to update the population in the algorithm's development phase; and implementing an optimal neighborhood perturbation strategy to determine the algorithm's termination conditions. When the conditions are met, the optimal solutions σ and γ are output; otherwise, the escape energy is recalculated and the population is continuously updated. The present invention utilizes adaptive weights and a variable spiral position update strategy to enhance the algorithm's global search capability, effectively avoiding the drawback of the traditional HHO algorithm, which is prone to falling into local optimal solutions during the search process.

Description

Technical Field

[0001] The present invention relates to the field of battery state of charge estimation, and in particular to a lithium battery SOC estimation algorithm, system, device and medium based on CSHHO-LSSVM (a least squares support vector machine hyperparameter optimization method based on a chaotic multi-strategy search Harris Hawk optimization algorithm). Background Art

[0002] In battery management systems (BMS), accurate estimation of the battery's state of charge (SOC) is crucial for extending battery life and ensuring battery safety. Traditional SOC estimation methods, such as the circuit-model-based Kalman filter (KF) and extended Kalman filter (EKF), typically rely on precise battery model parameters, but these parameters suffer from significant uncertainty in practical applications. To overcome this issue, machine learning methods have been gradually introduced into SOC estimation. Least Squares Support Vector Machines (LSSVMs), a commonly used nonlinear regression method, offer high estimation accuracy. However, the performance of LSSVMs is highly dependent on the choice of kernel function parameters (σ) and penalty parameter (γ). The optimization of these two hyperparameters is often considered a "black box" optimization problem, making it difficult to derive precise optimization strategies through theoretical analysis. Swarm intelligence optimization algorithms, particularly the Harris Hawk Optimizer (HHO), are widely used for LSSVM hyperparameter optimization due to their strong global search capabilities. However, traditional HHOs are prone to getting stuck in local optima during local search, compromising optimization performance. Therefore, improving HHO's global search capabilities is crucial for improving SOC estimation accuracy. Summary of the Invention

[0003] In view of the above-mentioned problems, the present invention is proposed.

[0004] Therefore, the problem to be solved by the present invention is: how to improve the global search capability of HHO, and thus improve the accuracy of SOC estimation.

[0005] To solve the above technical problems, the present invention provides the following technical solutions: a lithium battery SOC estimation algorithm based on CSHHO-LSSVM, which includes collecting SOC data of lithium phosphate batteries to form a sample set, dividing the sample set into two groups: training data and test data, using the training data to establish a mathematical model in simulation, and using the test data to verify the model; normalizing the sample set so that all sample data values are within the range of [0, 1]. The normalized conversion function is expressed as:

[0006] ,

[0007] in, is the normalized value, is the actual value, and Represent the maximum and minimum values respectively;

[0008] Set each initial parameter, including population size, maximum number of iterations, and dimension; initialize the population according to seven chaotic maps, where the independent variables consist of the kernel parameter σ and regularization parameter γ of the least squares vector machine, and calculate the individual fitness value; calculate the escape energy The formula is expressed as,

[0009] ,

[0010] in, Expressed as a random number (-1, 1); according to the position update formula, enter the exploration phase of the algorithm to update the population, introduce the variable spiral position update strategy to enter the development phase of the algorithm to update the population; use the optimal neighborhood perturbation strategy to execute the optimal neighborhood perturbation, judge the termination condition of the algorithm, when the condition is met, output the optimal solution σ and γ, otherwise recalculate the escape energy and continue to update the population; establish an LSSVM model based on the optimal solution σ and γ, use the obtained optimal estimation model to estimate the SOC of the power lithium battery, compare the estimated results with the measured results, and conduct targeted error analysis; the optimal neighborhood perturbation strategy is expressed as,

[0011] ,

[0012] in, and Represented as a random number uniformly generated between [0, 1], Indicates the current position. The new position generated; for the generated neighborhood position, a greedy strategy is used to decide whether to retain it. The calculation formula is expressed as,

[0013] ,

[0014] in, Expressed as Position fitness.

[0015] As a preferred solution of the lithium battery SOC estimation algorithm based on CSHHO-LSSVM described in the present invention, wherein: the seven chaotic maps include Sinusoidal chaotic map, Tent chaotic map, Kent chaotic map, Cubic chaotic map, Logistic chaotic map, Gauss chaotic map and Circle chaotic map; the initialization of the population according to the seven chaotic maps includes randomly generating in D-dimensional space Harris Hawk, the formula is expressed as,

[0016] ,

[0017] in, is represented as the Harris hawk population set, Represented as a Harris Hawk individual; each dimension of each Harris Hawk iterates M times, generating M Harris Hawks; after completing all Harris Hawk iterations, the seven chaotic maps are applied to the solution space and expressed as,

[0018] ,

[0019] in, Expressed as the upper bound of the d-dimensional exploration space, It is expressed as the lower bound of the d-dimensional exploration space; Expressed as the d-dimensional coordinates of the i-th Harris Hawk; It is represented as the coordinate of the ith Harris Hawk in the d-dimensional exploration space.

[0020] As a preferred solution of the lithium battery SOC estimation algorithm based on CSHHO-LSSVM described in the present invention, wherein: the method for constructing the position update formula includes forming an adaptive weight according to the current number of iterations t, improving the Harris Eagle optimization algorithm according to the adaptive weight, and obtaining the position update formula; the adaptive weight is formed according to the current number of iterations t. Expressed as,

[0021] ,

[0022] in, Expressed as the maximum number of iterations; the position update formula is expressed as,

[0023] ,

[0024] in, , , , , is represented as a random number in the range [0, 1), Denotes the optimal value of the i-th generation, Expressed as the average value of the population in generation i.

[0025] As a preferred solution of the lithium battery SOC estimation algorithm based on CSHHO-LSSVM described in the present invention, wherein: the introduction of the variable spiral position update strategy into the development phase of the algorithm to update the population includes introducing the variable spiral position update strategy into the position update formula, and updating the Harris Hawk population position as follows:

[0026] ,

[0027] in, Expressed as a control variable, it is designed according to the spiral mathematical model.

[0028] Another object of the present invention is to provide a lithium battery SOC estimation system based on CSHHO-LSSVM, which can find the optimal kernel function parameters of LSSVM in lithium battery SOC estimation.

[0029] To solve the above technical problems, the present invention provides the following technical solutions: a system for a lithium battery SOC estimation algorithm based on CSHHO-LSSVM, comprising: a data processing module, a parameter training module and an error analysis module; the data processing module collects the SOC data of the lithium phosphate battery to form a sample set, and normalizes the sample set data; the parameter training module optimizes the kernel parameter σ and regularization parameter γ of the least squares support vector machine based on the chaotic multi-strategy search Harris Eagle optimization algorithm, and outputs the obtained optimal solutions σ and γ; the error analysis module establishes an LSSVM model based on the optimal solutions σ and γ, uses the obtained optimal estimation model to estimate the SOC of the power lithium battery, and performs targeted error analysis based on the estimated results and the measured results.

[0030] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, it implements the steps of the above-mentioned lithium battery SOC estimation algorithm based on CSHHO-LSSVM.

[0031] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of a lithium battery SOC estimation algorithm based on CSHHO-LSSVM as described above.

[0032] The beneficial effects of the present invention are as follows: By combining the advantages of the swarm intelligence optimization algorithm and utilizing adaptive weights and variable spiral position update strategies, the method of the present invention enhances the global search capability of the algorithm, effectively avoiding the defect of the traditional HHO algorithm that is prone to falling into local optimal solutions during the search process. By introducing a chaotic mapping method to optimize population initialization, this method improves the coverage capability of the algorithm in the initial solution distribution stage, thereby further accelerating the convergence process of the optimal solution. Through this method, the SOC of the lithium battery can be estimated more accurately, and the effect of LSSVM hyperparameter optimization can be significantly improved, thereby improving the accuracy and stability of SOC estimation. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. 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 paying any creative work.

[0034] Figure 1 This is a flowchart of a Harris Hawk optimization algorithm based on chaotic multi-strategy search in Example 1.

[0035] Figure 2 This is a CSHHO optimization flowchart of a least squares support vector machine hyperparameter optimization method based on a chaotic multi-strategy search Harris Hawk optimization algorithm for lithium battery SOC estimation in Example 2.

[0036] Figure 3 This is a training set data diagram of the DST (sample 1) working condition of a least squares support vector machine hyperparameter optimization method based on a chaotic multi-strategy search Harris Eagle optimization algorithm for lithium battery SOC estimation in Example 2.

[0037] Figure 4 This is a training set data diagram of the FUDS (sample 2) working condition of a least squares support vector machine hyperparameter optimization method based on chaotic multi-strategy search Harris Eagle optimization algorithm for lithium battery SOC estimation in Example 2.

[0038] Figure 5 This is a training set data diagram of the BJDST (sample 3) working condition of a least squares support vector machine hyperparameter optimization method based on chaotic multi-strategy search Harris Hawk optimization algorithm for lithium battery SOC estimation in Example 2.

[0039] Figure 6 This is a test set data diagram of the US06 (sample 4) working condition of a least squares support vector machine hyperparameter optimization method based on chaotic multi-strategy search Harris Eagle optimization algorithm for lithium battery SOC estimation in Example 2.

[0040] Figure 7 This is an SOC prediction output diagram of a least squares support vector machine hyperparameter optimization method based on a chaotic multi-strategy search Harris Hawk optimization algorithm for lithium battery SOC estimation in Example 2.

[0041] Figure 8 This is a local magnified view of the SOC prediction error of a least squares support vector machine hyperparameter optimization method based on chaotic multi-strategy search Harris Hawk optimization algorithm for lithium battery SOC estimation in Example 2. DETAILED DESCRIPTION

[0042] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0043] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0044] Example 1

[0045] Reference Figure 1 This is the first embodiment of the present invention, which provides a Harris Hawk optimization algorithm based on chaotic multi-strategy search. The algorithm design principle is as follows: To effectively reduce the dependence of SOC estimation on circuit model parameters, the least squares support vector machine (LSSVM) is often used for SOC estimation. However, when using LSSVM to estimate lithium battery SOC, it is difficult to find the kernel function parameter σ and the penalty parameter γ. Theoretically, it is difficult to find a consistent relationship between the impact of these two parameters on the estimation accuracy of LSSVM. Therefore, optimizing the kernel function parameter σ and the penalty parameter γ can be considered a "black box" optimization problem. Swarm intelligence optimization algorithms can quickly obtain feasible solutions based solely on the input and output relationship without providing a functional expression for the optimization problem, making them an important method for solving black box optimization problems. The Harris Hawk Optimizer (HHO) algorithm is a typical swarm intelligence optimization algorithm. However, when optimizing LSSVM hyperparameters, it also suffers from relatively poor convergence and is prone to getting stuck in local optimal solutions. Therefore, this paper focuses on improving the HHO algorithm's search capabilities under different test functions, enhancing its ability to solve the "black box" optimization problem of finding LSSVM hyperparameters. The enhanced HHO algorithm is then used to search for LSSVM hyperparameters and estimate the SOC of lithium batteries.

[0046] By simulating the hunting process of Harris's hawks in nature, an adaptive weight and variable spiral position update strategy were introduced to enhance the algorithm's global search capability. In order to ensure that the initial solution generated by the HHO algorithm during the population initialization phase covers the solution space as much as possible, the optimal chaotic mapping method was selected from seven commonly used chaotic mapping population initialization methods and used as the population initialization method to improve the algorithm. In order to better find the algorithm's global optimal solution, an optimal neighborhood perturbation strategy was introduced. Therefore, based on the above scheme, the present invention designs a chaotic multi-strategy search Harris Hawk optimization algorithm (CSHHO) to improve the global search capability of the HHO algorithm and increase the speed at which HHO searches for the optimal solution.

[0047] Chaotic mapping: The traditional HHO algorithm adopts a random population initialization strategy. The limitation of this method is that it cannot ensure comprehensive coverage of the entire solution space. Chaotic sequences exhibit multiple advantages such as ergodicity, randomness, and regularity within a specific range, which means that it has the ability to thoroughly search the entire search space with a higher probability. Introducing chaotic sequences in the population initialization stage of the HHO algorithm can effectively enhance the global search capability of the algorithm, making it easier to escape from local optimal solutions and maintain the diversity of the population. The present invention selects seven common chaotic maps, namely Sinusoidal, Tent, Kent, Cubic, Logistic, Gauss, and Circle, for population initialization of the HHO algorithm. After analyzing the results, the optimal method for the HHO algorithm was selected as the population initialization method for the improved algorithm. The following are seven chaotic mappings:

[0048] Sinusoidal Chaotic Map

[0049] (1.1)

[0050] Where P is the control parameter , ;

[0051] Tent Chaos Map

[0052] (1.2)

[0053] Kent Chaos Map

[0054] (1.3)

[0055] Control parameters , when using chaos mapping, the initial value Cannot be compared with system parameters Otherwise, the system will evolve into a periodic system.

[0056] Cubic Chaos Map

[0057] (1.4)

[0058] Where b and c are the factors that influence the chaotic map. When b and c are different, the range of the Cubic chaotic map is also different. Take b = 4 and c = 3.

[0059] Logistic Chaotic Map

[0060] (1.5)

[0061] When P=4, the number of generated Logistic chaotic maps is between (0,1).

[0062] Gauss Chaos Map

[0063] (1.6)

[0064] Circle Chaos Map

[0065] (1.7)

[0066] like Figure 1 As shown in Figure 2, the three steps of using seven chaotic maps to initialize the population of the Harris Hawk Optimization algorithm HHO are:

[0067] Step 1: Generate randomly in D-dimensional space Harris Hawks, namely:

[0068] (1.8),

[0069] in, is represented as the Harris hawk population set, Represented as Harris's Hawk individual.

[0070] Step 2: Iterate each dimension of each Harris Hawk M times to generate M Harris Hawks.

[0071] Step 3: After all Harris Hawk iterations are completed, apply the chaos maps (1.1) to (1.7) to the solution space.

[0072] (1.9),

[0073] in, Expressed as the upper bound of the d-dimensional exploration space, It is expressed as the lower bound of the d-dimensional exploration space; Expressed as the d-dimensional coordinates of the i-th Harris Hawk; It is represented as the coordinate of the ith Harris Hawk in the d-dimensional exploration space.

[0074] The adaptive weight is formed according to the current number of iterations t. The Harris Eagle optimization algorithm is improved according to the adaptive weight, and the position update formula is obtained. Specifically, the adaptive weight composed of the current number of iterations t is as follows:

[0075] (1.10),

[0076] in, Expressed as the maximum number of iterations; this adaptive weight has the characteristic of nonlinear change between [0, 1], which is due to the cosine function in Therefore, at the beginning of the exploration phase, the weight value is small, but the change speed is slightly faster; at the end of the exploration phase, the weight value is large, but the change speed will slow down, thus fully ensuring the convergence of the algorithm. The improved HHO algorithm position update formula is:

[0077] (1.11),

[0078] in, , , , , is represented as a random number in the range [0, 1), Denotes the optimal value of the i-th generation, It is expressed as the average value of the population in the i-th generation. The position update after introducing the adaptive weight dynamically adjusts the weight size according to the increase in the number of iterations, so that the randomly selected Harris Hawk position and the optimal average Harris Hawk position in the population The guidance of Harris Hawk individuals is different at different times. As the number of iterations increases, the Harris Hawk population will move closer to the optimal position. A larger weight will speed up the movement of the Harris Hawk position, thereby accelerating the convergence of the algorithm.

[0079] A variable spiral position update strategy is introduced into the position update formula to update the position of the Harris Hawk population. Specifically, in order to simulate the actual process of prey searching in nature, a variable spiral position update strategy is introduced in the search phase of the HHO algorithm. In the exploration phase of the HHO algorithm, that is, in formula (1.11), the variable b is introduced to control the shape of the spiral, allowing the Harris Hawk to formulate more diverse search path strategies for position updates. The parameter b is designed to be a variable that changes with the number of iterations, dynamically adjusting the spiral shape of the Harris Hawk's exploration phase, enhancing the Harris Hawk's ability to explore unknown areas, and thus improving the algorithm's global search capability. It is defined as follows:

[0080] (1.12),

[0081] in, It is represented as a control variable, and the spiral shape is dynamically adjusted by introducing the number of iterations t based on the original spiral model. The parameters are designed so that the spiral shape changes from large to small as the number of iterations increases. In the early exploration phase of the HHO algorithm, the Harris Hawk searches for the target in a larger spiral shape. The Harris Hawk explores the global optimal solution as much as possible to improve the algorithm's global optimal search capability. In the later exploration phase of the HHO algorithm, the Harris Hawk searches for the target in a smaller spiral shape to improve the algorithm's search accuracy.

[0082] Introducing the optimal neighborhood perturbation strategy to find the global optimal solution of the algorithm Specifically, in order to improve the search efficiency of the HHO algorithm and avoid falling into the local optimum too early, the optimal neighborhood perturbation strategy is introduced in HHO. This random search not only increases the global search capability of the algorithm, but also prevents the algorithm from being premature to a certain extent. Specifically, in each iteration, a random perturbation is applied to the optimal position so that it can explore within the neighborhood. In this way, even if the current optimal position is close to the local optimum, the algorithm still has the opportunity to find a better global solution through neighborhood search. Neighborhood perturbation can be expressed as:

[0083] (1.13),

[0084] in, and, Represented as a random number uniformly generated between [0, 1], Indicates the current position. The new position generated. The size and direction of the perturbation are randomly determined to ensure the randomness and extensiveness of the search. This method can improve the convergence speed of the HHO algorithm and enhance its performance on global optimization problems.

[0085] For the generated neighborhood positions, a greedy strategy is used to decide whether to retain them. The calculation formula is:

[0086] (1.14),

[0087] in, yes The fitness of the position. If the generated position is better than the original position, it will be replaced with the original position, making it the global optimal position. Otherwise, the optimal position remains unchanged.

[0088] Impact of Chaos Strategy on HHO: To test and verify the performance of CSHHO, we designed two sets of experiments. Based on the classic 23 benchmark functions and the standard IEEE CEC 2017, all experiments are as follows:

[0089] Experiment 1: First, seven chaotic mapping methods were used as population initialization methods for HHO and tested separately. Second, the seven sets of data were analyzed and the optimal chaotic mapping method was selected as the population initialization method for the improved algorithm.

[0090] Experiment 2: First, building on Experiment 1, the HHO algorithm was tested by adding an adaptive weight mechanism, variable spiral position updates, and an optimal neighborhood perturbation mechanism. Secondly, the CSHHO algorithm was analyzed and compared with the HHO, WOA, SCA, and CSO algorithms. Third, the CSHHO algorithm was analyzed and compared with developed advanced variants such as IDSCA–HHOGCHHO, DEPSOASS, IGSA, and DGOBLFOA. Finally, building on the previous step, the algorithm-based accuracy scalability test was conducted using the IEEE CEC 2017 results, with test dimensions of D = 50 and D = 100.

[0091] To ensure the fairness of the experiment, the experiments were evaluated using the same parameters, with all population sizes N set to 30 and dimensions D set to 30; each algorithm was run 50 times independently on each test function. In each run, the function error value ,in is the average value of the test function over all iterations, is the best value of the test function recorded. In order to compare the performance of various algorithms, the mean error (Mean) and standard deviation (Std) of the function error value are used to evaluate the performance of the algorithm in all tests. Rank is also used to sort the average values of all algorithms in the simulation experiment from low to high. The lower the rank, the better the algorithm is compared with other algorithms; conversely, the higher the rank, the worse the algorithm is compared with other algorithms. Best records the best value in 50 experiments, and Worst records the worst value in 50 experiments. In addition, two non-parametric tests are designed, namely: the Wilcoxon signed rank test after Bonferroni-Holm correction is used to detect whether there is a significant performance difference between all algorithms; the Friedman test is used to rank the advantages and disadvantages of all algorithms. The test value is recorded in the ARV row, and Rank is its ranking.

[0092] To select the most effective chaotic mapping method among the classic chaotic mapping methods, enabling the algorithm to obtain the optimal initial solution position and accelerate the convergence of the Harris Hawk algorithm population, the sinusoidal chaotic mapping, tent chaotic mapping, Kent chaotic mapping, cubic chaotic mapping, logistic chaotic mapping, Gauss chaotic mapping, and circle chaotic mapping were selected to initialize the population of the HHO algorithm, forming the sinusoidal–HHO, tent–HHO, Kent–HHO, cubic–HHO, logistic–HHO, Gauss–HHO, and circle–HHO algorithms. The numerical performance of these seven algorithms was compared. Table 1 lists the results of these seven algorithms on 23 classic test functions. Table 2 shows the results of two nonparametric tests for these seven algorithms. The symbols "+\=\-" indicate the number of algorithms where the Gauss–HHO algorithm obtained a result that was greater, similar, or lower than that of the other algorithms. The ARV row in Table 2 shows the results of the Friedman test for the seven chaotic mapping HHO algorithms, and the Rank row shows their ranking.

[0093] Table 1 Comparison results of seven chaotic maps

[0094] ,

[0095] ,

[0096] ,

[0097] ,

[0098] Table 2 Results of Wilcoxon rank sum test and Friedman test after Bonferroni–Holm correction

[0099] ,

[0100] ,

[0101] By analyzing the experimental results, we conclude that under the UM functions (F1–F7), the Sinusoidal chaotic map achieves the best results in the F3, F5, and F7 test functions, while the Circle chaotic map achieves the best results in the F1 and F4 test functions. The Sinusoidal chaotic map has the greatest impact on the HHO algorithm, followed by the Gauss chaotic map and the Circle chaotic map. Under the MM functions (F8–F23), the Gauss chaotic map has the greatest impact on the HHO algorithm. The Circle chaotic map, the Sinusoidal chaotic map, the Tent chaotic map, and the Kent chaotic map achieve the best results in the F21, F15, F20, F13, and F23 test functions, respectively. The results of seven chaotic maps were compared across 23 test functions, and the Gauss chaotic map achieved the optimal solution.

[0102] Table 2 shows that among the 23 tested functions, the Gauss–HHO algorithm yielded superior results; the Friedman test results are presented in the ARV row. Gauss–HHO achieved the best average ranking, demonstrating that the population initialization of the HHO optimization algorithm using a Gaussian chaotic map also exhibits randomness, ergodicity, and initial value sensitivity. Using a Gaussian chaotic map to initialize the population of the HHO optimization algorithm yields better optimization accuracy.

[0103] Example 2

[0104] Reference Figure 2-Figure 8 , which is the second embodiment of the present invention, which is different from the first embodiment in that: a least squares support vector machine hyperparameter optimization method based on chaotic multi-strategy search Harris Hawk optimization algorithm for lithium battery SOC estimation includes:

[0105] In the proposed SOC estimation based on the CSHHO–LSSVM model, CSHHO is used to optimize the LSSVM algorithm parameters. The construction of the model includes the following parts, such as Figure 2 As shown:

[0106] (1) Sample collection. Collect the SOC data of lithium phosphate batteries to form a sample set, which serves as the training set for the SOC estimation model.

[0107] (2) Sample division. The samples are divided into two groups: training data and test data. In the simulation, the training data is used to build the mathematical model, and the test data is used to verify the model.

[0108] (3) Sample normalization. After normalization, all sample data values are in the range of [0, 1], which makes the model calculation faster. The normalized conversion function is,

[0109] (2.1),

[0110] in, is the normalized value, is the actual value, and Represent the maximum and minimum values respectively.

[0111] 4) Model construction. The steps of the SOC estimation algorithm based on the CSHHO–LSSVM model are as follows:

[0112] Step 21: Set each initial parameter, including population size, maximum number of iterations, and dimension.

[0113] Step 22: Initialize the population according to formulas (1.1-1.9), where the independent variable x consists of the kernel parameter σ and the regularization parameter γ of the least squares vector machine.

[0114] Step 23: Calculate individual fitness values.

[0115] Step 24: Calculate the escape energy E using the formula:

[0116] (2.2),

[0117] in, A random number represented as (-1, 1).

[0118] Step 25: Use formulas (1.10-1.11) to enter the exploration phase of the algorithm and update the population.

[0119] Step 26: Use formula (1.12) to enter the development phase of the algorithm to update the population.

[0120] Step 27: Use formulas (1.13-1.14) to perform the optimal neighborhood perturbation. Determine the algorithm termination condition. If the condition is met, output the optimal solutions σ and γ. Otherwise, continue to step 24.

[0121] Step 28: Establish an LSSVM model based on σ and γ. Use the obtained optimal estimation model to estimate the SOC of the power lithium battery. Compare the estimated results with the measured results and conduct targeted error analysis.

[0122] Performance indicators: In order to compare the performance differences between different models, this paper uses four evaluation indicators, namely root mean square error (RMSE), mean absolute error (MAE), mean absolute percentage error (MAPE), and coefficient of determination (R-squared, R2). The calculation formulas of the four evaluation indicators are:

[0123] (2.3),

[0124] (2.4),

[0125] (2.5),

[0126] (2.6),

[0127] in, is the SOC predicted value; is the true value of SOC; is the average SOC value; n is the number of predicted values. RMSE, MAE, and MAPE represent the fitting effect between the true value and the predicted value. The smaller the value, the more accurate the prediction result. It represents the degree of fit of the model to the data. The closer its value is to 1, the more accurate the prediction result is.

[0128] Experimental analysis: This simulation experiment uses a university 18650 battery data set to verify the proposed model, in which DST (sample 1), FUDS (sample 2), BJDST (sample 3) working condition data are selected as training sets, and US06 (sample 4) working condition is selected as test set. Figure 3-Figure 6 shown.

[0129] analyze Figure 7 It can be seen that under the same working conditions, when using CSHHO-LSSVM (a least squares support vector machine hyperparameter optimization method based on the chaotic multi-strategy search Harris Hawk optimization algorithm), GCHHO-LSSVM (a least squares support vector machine hyperparameter optimization method based on the Harris Hawk optimization algorithm with the introduction of Gaussian mutation and dimensional decision logic), and WOA-LSSVM (a least squares support vector machine hyperparameter optimization method based on the whale optimization algorithm) to predict SOC, the prediction output curve of CSHHO-LSSVM is closer to the expected value than the other methods, indicating that the prediction accuracy of CSHHO-LSSVM is higher than that of GCHHO-LSSVM and WOA-LSSVM. Figure 8As can be seen, under the US06 operating condition, the prediction errors of the proposed algorithms are generally maintained within 2%, while the prediction errors of the other algorithms are generally maintained within 16%. The training error shows that the CSHHO-LSSVM has more ideal SOC prediction results and smaller prediction errors.

[0130] Table 3 Comparison of SOC estimation results

[0131] ,

[0132] To further verify the accuracy of the prediction model, the CSHHO-LSSVM prediction results were compared with those of the GCHHO-LSSVM and WOA-LSSVM algorithms. The evaluation metrics for the battery SOC estimation results of the three models are shown in Table 3. It can be seen that the CSHHO-LSSVM algorithm of the present invention achieves RMSE reductions of approximately 0.13 and 0.43, MAPE% reductions of approximately 0.30 and 5.11, and MAE reductions of approximately 0.21 and 0.42, respectively, compared to the GCHHO-LSSVM and WOA-LSSVM algorithms, while R² increases of approximately 0.008 and 0.013, respectively, reflecting that the algorithm of the present invention has a smaller prediction error. In summary, the CSHHO-LSSVM model has higher accuracy and better regression fitting performance.

[0133] Example 3 is the third embodiment of the present invention, which is different from the previous two embodiments in that: a system for a least squares support vector machine hyperparameter optimization method based on a chaotic multi-strategy search Harris Hawk optimization algorithm for lithium battery SOC estimation, comprising a data processing module, a parameter training module and an error analysis module; the data processing module collects the SOC data of the lithium phosphate battery to form a sample set, and normalizes the sample set data; the parameter training module optimizes the kernel parameter σ and regularization parameter γ of the least squares support vector machine based on the chaotic multi-strategy search Harris Hawk optimization algorithm, and outputs the optimal solutions σ and γ; the error analysis module establishes an LSSVM model based on the optimal solutions σ and γ, uses the obtained optimal estimation model to estimate the SOC of the power lithium battery, and performs targeted error analysis based on the estimation results and the measured results.

[0134] If the functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or the portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include various media that can store program code, such as USB flash drives, mobile hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0135] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0136] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, and then editing, interpreting, or processing in another suitable manner as necessary, and then storing it in a computer memory.

[0137] It should be understood that various components of the present invention may be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods may be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof may be used: a discrete logic circuit having logic gate circuits for implementing logic functions on data signals, an application-specific integrated circuit having suitable combinational logic gate circuits, a programmable gate array (PGA), a field-programmable gate array (FPGA), etc.

[0138] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A lithium battery SOC estimation algorithm based on CSHHO-LSSVM, characterized by: include, Collect the SOC data of lithium phosphate batteries to form a sample set, which is divided into two groups: training data and test data. The training data is used to establish a mathematical model in the simulation, and the test data is used to verify the model. Normalize the sample set so that all sample data values are within the range of [0, 1]. The normalized conversion function is expressed as: ， in, is the normalized value, is the actual value, and Represent the maximum and minimum values respectively; Set each initial parameter, including population size, maximum number of iterations, and dimension; Initialize the population according to seven chaotic maps, where the independent variables are composed of the kernel parameter σ and the regularization parameter γ of the least squares vector machine, and calculate the individual fitness value; Calculating escape energy The formula is expressed as, ， in, A random number represented as (-1, 1), Represents the current iteration number, Expressed as the maximum number of iterations; According to the position update formula, enter the exploration phase of the algorithm to update the population, and introduce the variable spiral position update strategy to enter the development phase of the algorithm to update the population; The method for constructing the position update formula includes: Adaptive weights are formed, and the Harris Eagle optimization algorithm is improved according to the adaptive weights to obtain the position update formula; The current number of iterations Composition adaptive weight Expressed as, ， The position update formula is expressed as, ， in, , , , , is represented as a random number in the range [0, 1), Expressed as The optimal value of the iterative population, Expressed as The average value of the population of iterations, Represented as randomly selected Harris Hawk locations; The introduction of the variable spiral position update strategy into the development phase of the algorithm to update the population includes introducing the variable spiral position update strategy into the position update formula, and updating the Harris Hawk population position as follows: ， in, It is expressed as a control variable and designed according to the spiral mathematical model; The optimal neighborhood perturbation strategy is used to perform the optimal neighborhood perturbation and determine the termination condition of the algorithm. When the condition is met, the optimal solutions σ and γ are output. Otherwise, the escape energy is recalculated and the population is continued to be updated. Establish an LSSVM model based on the optimal solution σ and γ, use the obtained optimal estimation model to estimate the SOC of the power lithium battery, and perform targeted error analysis based on the comparison of the estimated results with the measured results; The optimal neighborhood perturbation strategy is expressed as, ， in, and Represented as a random number uniformly generated between [0, 1], Indicates the current position, Represented as the new position generated; For the generated neighborhood positions, a greedy strategy is used to decide whether to retain them. The calculation formula is expressed as: ， in, Expressed as Position fitness.

2. The lithium battery SOC estimation algorithm based on CSHHO-LSSVM according to claim 1, characterized in that: The seven chaotic maps include Sinusoidal chaotic map, Tent chaotic map, Kent chaotic map, Cubic chaotic map, Logistic chaotic map, Gauss chaotic map and Circle chaotic map; The initialization of the population according to the seven chaotic maps includes randomly generating Harris Hawk, the formula is expressed as, ， in, is represented as the Harris hawk population set, Represented as Harris's Hawk individuals; Each dimension of each Harris Hawk is iterated M times, generating M Harris Hawks; After completing all Harris Hawk iterations, seven types of chaos maps are applied to the solution space expressed as, ， in, Expressed as the upper bound of the d-dimensional exploration space, It is expressed as the lower bound of the d-dimensional exploration space; Expressed as the d-dimensional coordinates of the i-th Harris Hawk; It is represented as the coordinate of the ith Harris Hawk in the d-dimensional exploration space.

3. A system using the CSHHO-LSSVM-based lithium battery SOC estimation algorithm according to any one of claims 1 to 2, characterized in that: Including data processing module, parameter training module and error analysis module; The data processing module collects the SOC data of the lithium phosphate battery to form a sample set, and normalizes the sample set data; The parameter training module optimizes the kernel parameter σ and regularization parameter γ of the least squares support vector machine based on the chaotic multi-strategy search Harris Hawk optimization algorithm, and outputs the optimal solutions σ and γ; The error analysis module establishes an LSSVM model based on the optimal solutions σ and γ, uses the obtained optimal estimation model to estimate the SOC of the power lithium battery, and performs targeted error analysis based on the comparison of the estimated results with the measured results.

4. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of a lithium battery SOC estimation algorithm based on CSHHO-LSSVM according to any one of claims 1 to 2 are implemented.

5. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a lithium battery SOC estimation algorithm based on CSHHO-LSSVM according to any one of claims 1 to 2 are implemented.

Images