A lithium battery health state prediction method combining BWO optimization and GPR

Abstract

This invention discloses a lithium battery health state prediction method combining BWO optimization and GPR, belonging to the field of battery health monitoring. This method further improves the accuracy and robustness of predictions by combining the White Whale Optimization algorithm with a Gaussian Process Regression model. It can effectively handle input noise and complex data, and significantly improves the model's generalization ability and prediction accuracy through multi-model integration and secondary learning. It is suitable for various application scenarios, especially lithium battery management systems in mobile energy storage power supplies.

Description

Technical Field

[0001] This invention relates to the field of battery health monitoring technology, and in particular to a method for predicting the health status of lithium batteries by combining BWO optimization and GPR. Background Technology

[0002] Lithium-ion batteries, due to their high energy density, long lifespan, and good environmental adaptability, have been widely used in electric vehicles, mobile devices, and energy storage systems. To ensure safe battery use and extend their lifespan, accurately estimating the battery's State of Health (SOH) has become a key task in battery management systems (BMS). Currently, commonly used SOH estimation methods include prediction methods based on empirical models, equivalent circuit models, and data-driven models.

[0003] Methods based on empirical models and equivalent circuit models rely on large amounts of experimental data and complex electrochemical models, making them difficult to adapt to battery aging behavior under different operating conditions in practical applications. In contrast, data-driven models, with their advantages of not relying on physical models and being able to handle complex nonlinear relationships, are gradually becoming the mainstream research direction. Gaussian process regression (GPR), as a common machine learning model, is widely used in SOH estimation due to its ability to handle small sample data and its high prediction accuracy. However, traditional GPR models are prone to decreased prediction accuracy when faced with complex input noise. In addition, the selection of hyperparameters in existing technologies usually relies on manual parameter tuning or simple search algorithms, which cannot fully utilize optimization algorithms to improve model performance. Some researchers have proposed a GPR model based on genetic algorithm optimization to improve the accuracy of SOH estimation. However, existing optimization methods still suffer from insufficient generalization ability and poor adaptability when dealing with complex battery data. Summary of the Invention

[0004] This invention provides a lithium battery health state prediction method that combines BWO optimization and GPR, providing an efficient and reliable SOH prediction method, and offering strong support for the safe operation and extended service life of lithium batteries.

[0005] This invention provides a method for predicting the state of health of lithium batteries by combining BWO optimization and GPR, comprising the following steps:

[0006] Step 1: Obtain historical charge and discharge data of the lithium battery and preprocess the historical charge and discharge data;

[0007] Step 2: Randomly generate an initial beluga whale population, with each individual beluga whale corresponding to a set of hyperparameter combinations of the GPR model;

[0008] Step 3: Use the BWO algorithm to optimize the key hyperparameters of the GPR model, and maximize model performance by combining global search and random search mechanisms.

[0009] Step 4: Based on the optimized key hyperparameters, train the GPR model and the BWO-GPR model. Use the predicted outputs of the GPR model and the BWO-GPR model as input features to train the second-layer ensemble model. The second-layer model uses the GPR regression model to integrate the prediction results of the BWO-GPR model and the GPR model to form the final stacked ensemble model. The stacked ensemble model is used to predict the health status of the lithium battery.

[0010] Optionally, in one embodiment of the present invention, step 1 specifically includes:

[0011] Obtain historical charge and discharge data of lithium batteries, including historical voltage and current operating data;

[0012] The historical charge-discharge data is cleaned, normalized, and feature extracted to remove outliers and duplicates, supplement missing values, and convert features of different magnitudes to the same scale. Health features are then extracted: fixed voltage rise time, constant current charging time, and discharge efficiency. Fixed voltage rise time refers to the time required for a lithium battery to rise from one fixed voltage to another during charging; constant current charging time refers to the time required for the lithium battery to maintain a constant current during charging; and discharge efficiency refers to the ratio of the effective energy released by the lithium battery during discharge to its nominal capacity.

[0013] Optionally, in one embodiment of the present invention, step 2 specifically includes:

[0014] An initial beluga whale population is randomly generated, with each individual beluga whale having a population size of X. i Represents a group The value represents a set of hyperparameters for each beluga whale individual, corresponding to a set of length scales l and signal variances in the GPR model.

[0015] In the GPR model, the kernel function is used to define the similarity between data points. The expression for the radial basis function kernel is:

[0016]

[0017] Where, x i For the i-th data point, x j Let j be the j-th data point.

[0018] Optionally, in one embodiment of the present invention, step 3 specifically includes:

[0019] (1) Fitness assessment: The fitness value of each beluga whale individual was calculated, and the model performance was measured by maximizing the log marginal likelihood function;

[0020] (2) Location Update: The beluga whale's location is updated using the standard beluga whale foraging behavior and Levy flight random search in the BWO algorithm, so that the fitness value of the new location gradually increases. The specific location update formula is as follows:

[0021] X i (t+1)=X i (t)+α(X best (t)-X i (t))+β·Levy(λ)

[0022] Among them, X i (t) represents the position of the i-th beluga whale in the t-th iteration, i.e., the current hyperparameter combination; X best (t) represents the current globally optimal hyperparameter combination; α and β are parameters controlling the step size; Levy(λ) is the Levy flight function used for random search;

[0023] (3) Fitness assessment and selection: For the newly generated position, i.e. the new hyperparameter combination, calculate the fitness value, retain the beluga whale individuals with higher fitness values ​​and enter the next generation of iteration process;

[0024] (4) Iterate until convergence: Continue iterating until the predetermined maximum number of iterations T is reached. max Alternatively, if the change in fitness value is lower than a preset threshold, the optimal combination of hyperparameters can be obtained.

[0025] The BWO algorithm transitions from the exploration phase to the development phase based on the balance coefficient Bf, and its model is as follows:

[0026] B f =B0(1-T / 2T) max )

[0027] Where B0 is the randomly varying fundamental coefficient, and T is the current iteration number. max To determine the maximum number of iterations, B0 changes randomly between (0,1) in each iteration. When the balance coefficient B... f When the equilibrium coefficient B is greater than 0.5, the search phase begins. f When B is ≤0.5, the development phase begins. As the number of iterations T increases, B... f The fluctuation range decreased from (0,1) to (0,0.5), indicating that the probability changed significantly from the search stage to the development stage, and the probability of the development stage increased with the continuous increase of the number of iterations T.

[0028] This invention presents a lithium-ion battery state-of-health (SOH) prediction method combining Beluga Whale Optimization (BWO) and Gaussian Process Regression (GPR). By integrating the BWO algorithm and a Gaussian Process Regression (GPR) model and employing a stacked ensemble secondary learning approach, it achieves high-precision prediction of the SOH of lithium-ion batteries. The BWO-optimized GPR model effectively addresses the weaknesses of traditional models in handling noisy data and the difficulty in selecting hyperparameters, improving the model's adaptability and robustness. Furthermore, the stacked ensemble strategy, through multi-model fusion, further enhances the accuracy and robustness of the prediction results. Therefore, this invention not only provides reliable SOH estimation in various application scenarios but also effectively improves the safety and efficiency of lithium-ion battery management systems.

[0029] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0030] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0031] Figure 1 A flowchart of a lithium battery health state prediction method combining BWO optimization and GPR provided according to an embodiment of the present invention;

[0032] Figure 2 This is a comparison chart of the predicted SOH curve and the actual SOH curve in an embodiment of the present invention. Detailed Implementation

[0033] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0034] Existing methods for estimating the state of health (SQH) of lithium batteries are mainly divided into two categories: physical model-based and data-driven. However, physical model-based methods rely on complex electrochemical and equivalent circuit models. Although their physical meaning is clear, they have poor adaptability to the diverse usage environments of different battery types and struggle to meet accuracy requirements in practical applications. On the other hand, while data-driven methods excel in handling complex battery data, most existing methods still fall short in handling noisy data, improving model generalization ability, and real-time prediction performance, making them difficult to adapt to the complex and ever-changing usage scenarios of lithium batteries. Furthermore, the selection and optimization of model hyperparameters still rely on manual parameter tuning or simple search algorithms, affecting the accuracy and efficiency of prediction. To address these issues, this invention provides a lithium battery SQH prediction method combining BWO optimization and GPR. By combining the White Whale optimization algorithm with a Gaussian process regression model, the accuracy and robustness of the prediction are further improved. This method can effectively handle input noise and complex data, and significantly improves the model's generalization ability and prediction accuracy through multi-model ensemble and secondary learning. It is suitable for various application scenarios, especially lithium battery management systems in mobile energy storage power supplies. By optimizing the model's hyperparameters and improving its adaptability, an efficient and reliable method for predicting state of oxygen (SOH) is provided, offering strong support for the safe operation and extended lifespan of lithium batteries.

[0035] This invention proposes a lithium battery SOH estimation model based on the integration of BWO-optimized GPR and traditional GPR stacking. This method combines the global search capability of the White Whale Optimization Algorithm (BWO) with the small sample data processing capability of the GPR model, significantly improving the accuracy and robustness of lithium battery SOH estimation. The main steps are as follows:

[0036] Step 1: Obtain historical charge and discharge data of the lithium battery and preprocess the historical charge and discharge data.

[0037] First, data preparation: Collect historical charging and discharging data of lithium batteries, including voltage and current operating data. Second, data preprocessing: Clean, normalize, and extract features from the data. This involves removing outliers and duplicates, filling in missing values, converting features of different magnitudes to the same scale, and extracting healthy features: fixed voltage rise time, constant current charging time, and discharge efficiency. Fixed voltage rise time refers to the time required for the battery to rise from one fixed voltage to another during charging; constant current charging time refers to the time required for the battery to maintain a constant current during charging; and discharge efficiency refers to the ratio of the effective energy released by the battery during discharge to its nominal capacity.

[0038] Step 2: Randomly generate an initial beluga whale population, with each individual beluga whale corresponding to a set of hyperparameter combinations of the GPR model.

[0039] In an embodiment of the present invention, step 2 specifically includes:

[0040] An initial beluga whale population is randomly generated, with each individual beluga whale having a population size of X. i Represents a group The value represents a set of hyperparameters for each beluga whale individual, corresponding to a set of length scales l and signal variances in the GPR model.

[0041] In the GPR model, the kernel function is used to define the similarity between data points. One common form of kernel function is the Radial Basis Function (RBF kernel), which is expressed as follows:

[0042]

[0043] Where, x i For the i-th data point, x j Let j be the j-th data point.

[0044] Step 3: Use the BWO algorithm to optimize the key hyperparameters of the GPR model, and maximize model performance by combining global search and random search mechanisms.

[0045] The BWO algorithm is used to optimize the key hyperparameters of the GPR model. By combining global search and random search mechanisms, the hyperparameter space is effectively explored to maximize model performance.

[0046] In an embodiment of the present invention, step 3 specifically includes:

[0047] (1) Fitness assessment: The fitness value of each beluga whale individual was calculated, and the model performance was measured by maximizing the log marginal likelihood function;

[0048] (2) Location Update: The beluga whale's location is updated using the standard beluga whale foraging behavior and Levy flight random search in the BWO algorithm, so that the fitness value of the new location gradually increases. The specific location update formula is as follows:

[0049] X i (t+1)=X i (t)+α(X best (t)-X i (t))+β·Levy(λ)

[0050] Among them, X i (t) represents the position of the i-th beluga whale in the t-th iteration, i.e., the current hyperparameter combination; X best (t) represents the current globally optimal hyperparameter combination; α and β are parameters controlling the step size; Levy(λ) is the Levy flight function used for random search;

[0051] (3) Fitness assessment and selection: For the newly generated position, i.e. the new hyperparameter combination, calculate the fitness value, retain the beluga whale individuals with higher fitness values ​​and enter the next generation of iteration process;

[0052] (4) Iterate until convergence: Continue iterating until the predetermined maximum number of iterations T is reached. max Alternatively, if the change in fitness value is lower than a preset threshold, the optimal combination of hyperparameters can be obtained.

[0053] The BWO algorithm transitions from the exploration phase to the development phase based on the balance coefficient Bf, and its model is as follows:

[0054] B f =B0(1-T / 2T) max )

[0055] Where B0 is the randomly varying fundamental coefficient, and T is the current iteration number. max To determine the maximum number of iterations, B0 changes randomly between (0,1) in each iteration. When the balance coefficient B... f When the equilibrium coefficient B is greater than 0.5, the search phase begins. f When B is ≤0.5, the development phase begins. As the number of iterations T increases, B... f The fluctuation range decreased from (0,1) to (0,0.5), indicating that the probability changed significantly from the search stage to the development stage, and the probability of the development stage increased with the continuous increase of the number of iterations T.

[0056] Step 4: Based on the optimized key hyperparameters, train the GPR model and the BWO-GPR model. Use the predicted outputs of the GPR model and the BWO-GPR model as input features to train the second-layer ensemble model. The second-layer model uses the GPR regression model to integrate the prediction results of the BWO-GPR model and the GPR model to form the final stacked ensemble model. The stacked ensemble model is used to predict the health status of the lithium battery.

[0057] The following detailed description of the lithium battery health status prediction method combining BWO optimization and GPR of the present invention will be provided through specific embodiments.

[0058] exist Figure 1The process begins with data acquisition and preprocessing, followed by data splitting into training and test sets. Next, two different GPR models are trained: a BWO-optimized GPR model and a standard GPR model. The outputs of these two models are then used as input to a second-layer GPR model for further learning. Finally, model performance is evaluated on the test set, and the lithium battery's state of health (SOH) estimate is output. This flowchart clearly illustrates the complete steps from data processing to model training and result output, demonstrating the core idea of ​​using stacked ensemble and optimization algorithms for SOH prediction.

[0059] The data collected from the cyclic operation of a 179Ah CBOMHW3NA battery at 25℃ was used to train the model, and the remaining 50% was used to validate the model.

[0060] The parameters of the lithium batteries studied are shown in Table 1. These parameters include battery capacity, voltage range, cycle life, and operating temperature range, which are important bases for experimental design and result analysis.

[0061] Table 1 Lithium Battery Parameters

[0062]

[0063] from Figure 2 As can be seen, the proposed stacked integrated model performs exceptionally well in predicting the state of health (SOH) of lithium-ion batteries. Comparison reveals that the predicted curve of this invention closely matches the actual SOH curve, especially in the early and middle stages of battery degradation, where the model's predicted values ​​almost perfectly match the actual values. This indicates that the model has high accuracy in capturing battery degradation trends. Furthermore, the shaded areas in the figure represent the confidence intervals of the prediction results. It can be observed that the narrow confidence intervals indicate that the model not only accurately predicts the battery's SOH value but also demonstrates good control over prediction uncertainty. This is particularly important because, in practical applications, prediction uncertainty directly affects the decision-making process of the battery management system (BMS). The reasonable range of the confidence intervals further demonstrates the robustness and reliability of this model under different operating conditions.

[0064] In addition, based on evaluation metrics, the root mean square error (RMSE) of the model in this invention is 0.0021809, and the mean absolute error (MAE) is 0.0016394, indicating that the model's estimation error for the state of harmonics (SOH) of lithium batteries is extremely small, demonstrating very high accuracy. Furthermore, the coefficient of determination (R²) is also very high. 2The value reached 0.99483, close to 1, indicating that the model has excellent fitting performance and can accurately capture the changing trend of SOH. This shows that the model can maintain high prediction accuracy under different conditions and has strong generalization ability and robustness.

[0065] The lithium battery health state prediction method proposed in this embodiment of the invention combines BWO optimization and GPR, further improving the accuracy and robustness of predictions by integrating the White Whale Optimization Algorithm with a Gaussian Process Regression Model. This method effectively handles input noise and complex data, and significantly improves the model's generalization ability and prediction accuracy through multi-model ensemble and secondary learning. It is suitable for various application scenarios, especially lithium battery management systems in mobile energy storage power supplies.

[0066] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0067] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0068] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more N executable instructions for implementing custom logic functions or processes, and the scope of preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of the invention pertain.

Claims

1. A method for predicting the state of health of lithium batteries by combining BWO optimization and GPR, characterized in that, Includes the following steps: Step 1: Obtain historical charge and discharge data of the lithium battery and preprocess the historical charge and discharge data; Step 2: Randomly generate an initial beluga whale population, with each beluga whale individual corresponding to a set of hyperparameter combinations of the GPR model; Step 3: Use the BWO algorithm to optimize the key hyperparameters of the GPR model, and maximize model performance by combining global search and random search mechanisms. Step 4: Based on the optimized key hyperparameters, train the GPR model and the BWO-GPR model. Use the predicted outputs of the GPR model and the BWO-GPR model as input features to train the second-layer ensemble model. The second-layer model uses the GPR regression model to integrate the prediction results of the BWO-GPR model and the GPR model to form the final stacked ensemble model. The stacked ensemble model is used to predict the health status of the lithium battery. Step 3 specifically includes: (1) Fitness assessment: The fitness value of each beluga whale individual was calculated, and the model performance was measured by maximizing the log marginal likelihood function; (2) Location Update: The beluga whale's location is updated using the standard beluga whale foraging behavior and Levy flight random search in the BWO algorithm, so that the fitness value of the new location gradually increases. The specific location update formula is as follows: X i (t+1)=X i (t)+α(X best (t)-X i (t))+β·Levy(λ) Among them, X i (t) represents the position of the i-th beluga whale in the t-th iteration, i.e., the current hyperparameter combination; X best (t) represents the current globally optimal hyperparameter combination; α and β are parameters controlling the step size; Levy(λ) is the Levy flight function used for random search; (3) Fitness assessment and selection: For the newly generated position, i.e. the new hyperparameter combination, calculate the fitness value, retain the beluga whale individuals with higher fitness values ​​and enter the next generation of iteration process; (4) Iterate until convergence: Continue iterating until the predetermined maximum number of iterations T is reached. max Alternatively, if the change in fitness value is lower than a preset threshold, the optimal combination of hyperparameters can be obtained. The BWO algorithm transitions from the exploration phase to the development phase based on the balance coefficient Bf, and its model is as follows: B f =B0(1-T / 2T max ) Where B0 is the randomly varying fundamental coefficient, and T is the current iteration number. max To determine the maximum number of iterations, B0 changes randomly between (0,1) in each iteration. When the balance coefficient B... f When the equilibrium coefficient B is greater than 0.5, the search phase begins. f When B is ≤0.5, the development phase begins. As the number of iterations T increases, B... f The fluctuation range decreased from (0,1) to (0,0.5), indicating that the probability changed significantly from the search stage to the development stage, and the probability of the development stage increased with the continuous increase of the number of iterations T.

2. The method according to claim 1, characterized in that, Step 1 specifically includes: Obtain historical charge and discharge data of lithium batteries, including historical voltage and current operating data; The historical charge-discharge data is cleaned, normalized, and feature extracted to remove outliers and duplicates, supplement missing values, and convert features of different magnitudes to the same scale. Health features are then extracted: fixed voltage rise time, constant current charging time, and discharge efficiency. Fixed voltage rise time refers to the time required for a lithium battery to rise from one fixed voltage to another during charging; constant current charging time refers to the time required for the lithium battery to maintain a constant current during charging; and discharge efficiency refers to the ratio of the effective energy released by the lithium battery during discharge to its nominal capacity.

3. The method according to claim 1, characterized in that, Step 2 specifically includes: An initial beluga whale population is randomly generated, with each individual beluga whale having a population size of X. i Represents a group The value represents a set of hyperparameters for each beluga whale individual, corresponding to a set of length scales l and signal variances in the GPR model. In the GPR model, the kernel function is used to define the similarity between data points. The expression for the radial basis function kernel is: Where, x i For the i-th data point, x j Let j be the j-th data point.

Images