Battery RUL prediction method, device, equipment and medium based on variable mode decomposition

Abstract

The present application discloses a battery RUL prediction method, apparatus, device and medium based on variable mode decomposition, which relates to the field of battery technology, including: obtaining battery capacity degradation data and battery capacity failure threshold, using variable mode decomposition to smooth the battery capacity degradation data to obtain multiple modal components, obtaining battery capacity key feature data based on the multiple modal components, and obtaining a battery capacity prediction value by inputting the battery capacity key feature data into an optimized support vector regression model. When the battery capacity prediction value is less than the battery capacity failure threshold, the remaining battery service life is calculated. By using variable mode decomposition to smooth the battery capacity degradation data, extracting key features and inputting them into the optimized SVR model to predict the battery capacity, and when the predicted battery capacity is lower than the failure threshold, the remaining battery service life is calculated, thereby improving the RUL prediction accuracy and robustness.

Description

Technical Field

[0001] The present invention relates to the field of battery technology, and in particular to a battery RUL prediction method, device, equipment and medium based on variable mode decomposition. Background Art

[0002] Lithium-ion batteries are widely used in consumer electronics, electric vehicles, and space systems due to their higher energy density, longer cycle life, and lower self-discharge rate. However, over time, the performance of lithium-ion batteries gradually degrades, an irreversible change that affects battery safety and reliability.

[0003] Support vector regression (SVR) is a commonly used method for lithium battery RUL prediction. Traditional SVR methods are limited in their generalization capabilities when faced with high data noise or large fluctuations in nonlinear degradation processes, resulting in reduced prediction accuracy. Relying solely on PSO for parameter optimization also has certain limitations, such as being prone to falling into local optimal solutions. To address these issues, an optimization scheme combining variable mode decomposition is proposed to improve the accuracy of SVR model predictions. Summary of the Invention

[0004] The main purpose of this application is to provide a battery RUL prediction method, device, equipment and medium based on variable mode decomposition, aiming to solve the technical problem of how to improve the accuracy of battery RUL prediction.

[0005] To achieve the above objectives, the present application proposes a battery RUL prediction method based on variable mode decomposition, the method comprising:

[0006] Obtain battery capacity degradation data and battery capacity failure threshold;

[0007] Smoothing the battery capacity degradation data using variable mode decomposition to obtain multiple modal components;

[0008] Obtaining key characteristic data of battery capacity according to the plurality of modal components;

[0009] Obtaining a predicted value of battery capacity by inputting the key characteristic data of battery capacity into an optimized support vector regression model;

[0010] When the battery capacity prediction value is less than the battery capacity failure threshold, the remaining battery service life is calculated.

[0011] In one embodiment, before the step of inputting the key characteristic data of the battery capacity into the optimized support vector regression model to obtain the predicted value of the battery capacity, the step includes:

[0012] Dividing the battery capacity key feature data into a training set and a test set;

[0013] Obtain training set feature data and establish an initial support vector regression model;

[0014] The initial support vector regression model is optimized according to the particle swarm optimization algorithm based on the training set feature data to obtain an optimized support vector regression model.

[0015] In one embodiment, before the step of optimizing the initial support vector regression model according to the training set feature data using a particle swarm optimization algorithm to obtain an optimized support vector regression model, the step includes:

[0016] Obtain particle filtering in the training set feature data;

[0017] Initialize algorithm parameters and set category measurement thresholds;

[0018] Dividing the particle population in the particle filter into multiple subgroups;

[0019] Calculate the fitness value of each particle in each subgroup, and initialize the best position of each subgroup individual and the best position of the group;

[0020] Update the fitness value of each subgroup individual, the best position of each subgroup individual and the best position of the group according to the functional relationship;

[0021] Determine whether the type metric value exceeds the type metric threshold. If the particle type metric value is less than the type metric threshold, re-plan the particle population.

[0022] Determine whether the maximum number of iterations has been reached. If so, output the model parameters.

[0023] After the model parameters are input into the nonlinear fitting function, the parameter penalty factor and kernel parameters for constructing the optimized support vector model are obtained.

[0024] In one embodiment, the step of optimizing the initial support vector regression model according to the training set feature data using a particle swarm optimization algorithm to obtain an optimized support vector regression model includes:

[0025] Obtaining a particle swarm from the training set data information;

[0026] Initializing the positions and velocities of particles in the particle swarm and calculating the fitness of the particles;

[0027] Set the number of iterations and calculate the historical optimal value of the particle;

[0028] By comparing the fitness of the particles in each iteration, a global optimal value of the particle swarm is obtained;

[0029] updating the position and velocity of the particle according to the historical optimal value and the global optimal value;

[0030] Determine whether the position and velocity of the particle meet a preset range, and if not, perform boundary condition processing on the particle;

[0031] If the number of iterations is reached or the error is less than the preset error, the optimized support vector regression model is obtained.

[0032] In one embodiment, the step of obtaining a battery capacity prediction value by inputting the battery capacity key feature data into an optimized support vector regression model includes:

[0033] Get the preset starting point cycle period and the selected cycle period;

[0034] When the selected cycle period exceeds the cycle period of the preset starting point, the selected cycle period is substituted into the optimized support vector regression model to obtain a predicted value of battery capacity.

[0035] In one embodiment, the step of smoothing the battery capacity degradation data using variable mode decomposition to obtain multiple modal components includes:

[0036] Get modal decomposition parameters;

[0037] According to the modal decomposition parameters, the number of modal components and the center frequencies of the modal components are obtained;

[0038] The battery capacity degradation data is iteratively decomposed based on the number of modal components and the center frequency to extract multiple modal components.

[0039] In one embodiment, when the battery capacity prediction value is less than the battery capacity failure threshold, the step of calculating the remaining battery service life further includes:

[0040] Get preset period and select cycle period;

[0041] When the battery capacity prediction value is greater than the battery capacity failure threshold, the selected cycle period is added to the preset cycle period as a new cycle period to obtain the battery capacity prediction value of the new cycle period, and so on, until the battery capacity prediction value of the new cycle period does not exceed the battery capacity failure threshold.

[0042] In addition, to achieve the above objectives, the present application also proposes a battery RUL prediction device based on variable mode decomposition, the battery RUL prediction device based on variable mode decomposition comprising:

[0043] An acquisition module, used to obtain battery capacity degradation data and battery capacity failure threshold;

[0044] a processing module, configured to smooth the battery capacity degradation data using variable modal decomposition to obtain a plurality of modal components;

[0045] an extraction module, configured to obtain key characteristic data of battery capacity according to the plurality of modal components;

[0046] a calculation module, configured to obtain a predicted value of battery capacity by inputting the key characteristic data of battery capacity into an optimized support vector regression model;

[0047] The result module is used to calculate the remaining service life of the battery when the battery capacity prediction value is less than the battery capacity failure threshold.

[0048] In addition, to achieve the above-mentioned purpose, the present application also proposes a medium, which is a computer-readable medium and stores a computer program. When the computer program is executed by a processor, the steps of the battery RUL prediction method based on variable modal decomposition as described above are implemented.

[0049] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the steps of the battery RUL prediction method based on variable mode decomposition as described above.

[0050] This application obtains battery capacity degradation data and battery capacity failure threshold, uses variable mode decomposition to smooth the battery capacity degradation data, obtains multiple modal components, obtains key feature data of battery capacity based on the multiple modal components, and inputs the key feature data of battery capacity into an optimized support vector regression model to obtain a battery capacity prediction value. When the battery capacity prediction value is less than the battery capacity failure threshold, the remaining battery service life is calculated. By using variable mode decomposition to smooth the battery capacity degradation data, extracting key features and inputting them into the optimized SVR model to predict the battery capacity, when the predicted battery capacity is lower than the failure threshold, the remaining battery service life is calculated, thereby improving the RUL prediction accuracy and robustness. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0052] Figure 1This is a flow chart of the first embodiment of the battery RUL prediction method based on variable mode decomposition of this application;

[0053] Figure 2 This is a battery parameter table diagram of the first embodiment of the battery RUL prediction method based on variable mode decomposition of this application;

[0054] Figure 3 This is an A01 battery RUL prediction performance evaluation diagram of the first embodiment of the battery RUL prediction method based on variable mode decomposition of this application;

[0055] Figure 4 This is an A02 battery RUL prediction performance evaluation diagram of the first embodiment of the battery RUL prediction method based on variable mode decomposition of this application;

[0056] Figure 5 This is the A01 lithium battery RUL diagram for early and late battery degradation prediction of the first embodiment of the battery RUL prediction method based on variable mode decomposition of this application;

[0057] Figure 6 This is the A02 lithium battery RUL diagram for early and late battery degradation prediction of the first embodiment of the battery RUL prediction method based on variable mode decomposition of this application;

[0058] Figure 7 This is the APSO and PSO fitness curve diagram of the first embodiment of the battery RUL prediction method based on variable mode decomposition of this application;

[0059] Figure 8 This is a flow chart of the second embodiment of the battery RUL prediction method based on variable mode decomposition of this application;

[0060] Figure 9 This is a flow chart of the third embodiment of the battery RUL prediction method based on variable mode decomposition of this application;

[0061] Figure 10 This is a schematic diagram of the module structure of a battery RUL prediction device based on variable mode decomposition according to an embodiment of the present application;

[0062] Figure 11 Schematic diagram of the device structure of the hardware operating environment involved in the battery RUL prediction method based on variable mode decomposition in an embodiment of the present application.

[0063] The purpose, features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0064] It should be understood that the specific embodiments described herein are merely used to explain the technical solutions of the present application and are not intended to limit the present application.

[0065] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0066] The main solution of the embodiment of the present application is: by obtaining battery capacity degradation data and battery capacity failure threshold, using variable mode decomposition to smooth the battery capacity degradation data to obtain multiple modal components, obtaining battery capacity key feature data based on the multiple modal components, and inputting the battery capacity key feature data into the optimized support vector regression model to obtain the battery capacity prediction value. When the battery capacity prediction value is less than the battery capacity failure threshold, the remaining battery service life is calculated.

[0067] Based on this, the embodiment of the present application provides a battery RUL prediction method based on variable mode decomposition, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the battery RUL prediction method based on variable mode decomposition of this application.

[0068] In this embodiment, the battery RUL prediction method based on variable mode decomposition includes steps S10 to S50:

[0069] Step S10: Obtain battery capacity degradation data and a battery capacity failure threshold.

[0070] It's important to note that to obtain reliable battery capacity degradation data, rigorously controlled cycle life experiments are conducted to record the battery capacity after each charge and discharge cycle, creating a capacity degradation curve over time. This curve illustrates how battery capacity decreases with increasing cycle number and reflects aging trends under different conditions. For batteries in real-world applications, due to the complex and variable operating environments influenced by factors such as temperature, load pattern, and charge rate, sensors or specialized data acquisition equipment are required to monitor and accumulate this data over a long period of time. The standard for determining battery capacity failure thresholds is that a battery is considered to have reached end-of-life (EOL) when its capacity drops to approximately 80% of its initial nominal capacity. This standard is based on considerations of battery performance and safety. Below this threshold, the battery no longer meets application requirements and will experience more rapid performance degradation. However, in some specialized applications, such as electric vehicles and aerospace, different failure thresholds are set to meet higher safety requirements or optimize cost-effectiveness.

[0071] Specifically, in this embodiment, the experimental battery capacity degradation data comes from 8 lithium batteries (battery data sets A01, A02, A03, A04 provided by Stanford University and battery data sets cs2_35, cs2_36, cs2_37, cs2_38 from CACLE of the University of Maryland).

[0072] Step S20 : ​​Smoothing the battery capacity degradation data using variable modal decomposition to obtain multiple modal components.

[0073] It should be noted that Variational Mode Decomposition (VMD) is an advanced algorithm for processing nonlinear and non-stationary signals. It decomposes complex battery capacity degradation data into multiple intrinsic mode functions (IMFs) with different center frequencies. Each IMF represents a characteristic feature of the original data. The process of smoothing battery capacity degradation data using VMD begins with formulating an optimization problem. The goal is to find an optimal set of IMFs that minimize their bandwidth and whose sum equals the original signal. VMD solves this constrained optimization problem by introducing a quadratic penalty factor and Lagrange multipliers to ensure that there is no cross-influence between the IMFs. Specifically, for each IMF, its Hilbert transform associated with the instantaneous phase is calculated and modulated to baseband to obtain spectral information. The bandwidth of the IMF is then estimated. Next, the IMF and its center frequency are iteratively updated using the alternating direction multiplication method until convergence is met. This results in multiple noise-free and independent IMF components.

[0074] Step S30: obtaining key characteristic data of battery capacity according to multiple modal components.

[0075] It's important to note that from the multiple IMFs described above, we screen out those components most relevant to battery capacity decay trends, such as low-frequency IMFs that exhibit long-term, slow variations. This can be accomplished by analyzing the Pearson correlation coefficients between the individual IMFs, selecting those with strong or extremely strong correlations as the basis for reconstructing the lithium battery capacity data. This not only eliminates high-frequency noise interference but also preserves the key dynamic characteristics of the original data. Finally, based on these processed IMF components, we use an optimized support vector regression (APSO-SVR) model to predict the remaining battery life.

[0076] Step S40 , obtaining a predicted value of battery capacity by inputting key characteristic data of battery capacity into an optimized support vector regression model.

[0077] It should be noted that before using the optimized support vector model for prediction, two important parameters need to be determined: the penalty factor and the kernel parameter. These two parameters determine the complexity of the APSO-SVR model and the degree of fit to the training data.

[0078] Furthermore, once the optimization is complete and the parameters are selected, the next step is to feed the key characteristic data of the battery capacity obtained through VMD processing into the APSO-SVR model as input variables. These characteristic data contain important information reflecting the battery aging process, such as the capacity decay rate and the degradation pattern under the influence of temperature. By learning these features, the APSO-SVR model can capture the intrinsic connection between battery capacity changes and time, and then output a predicted value of the battery capacity. It is worth noting that in order to further improve the prediction effect, other relevant factors can also be considered as auxiliary inputs, such as ambient temperature, charge and discharge current, etc., which will help to more comprehensively describe the working status of the battery.

[0079] Step S50: When the predicted battery capacity value is less than the battery capacity failure threshold, the remaining battery life is calculated.

[0080] It's important to note that when the predicted battery capacity falls below the preset failure threshold, it means the lithium-ion battery may be nearing the end of its life. To ensure safe system operation and avoid unexpected failures, the battery's remaining useful life (RUL) needs to be calculated. The RUL is the time from the current moment until the battery completely fails.

[0081] Specifically, the previously established APSO-SVR model is used to predict future battery capacity trends. By inputting the latest battery status information into the APSO-SVR model, a capacity prediction curve for the future period is obtained. The difference between this predicted curve and the failure threshold is then used to estimate the battery's remaining operating life, or RUL.

[0082] Furthermore, the preset starting point cycle period and the selected cycle period are obtained. When the selected cycle period exceeds the cycle period of the preset starting point, the selected cycle period is substituted into the optimized support vector regression model to obtain the battery capacity prediction value. Specifically, the preset starting point cycle period refers to the reference number of cycles used to start RUL prediction, which is usually determined based on when the battery capacity decays to a certain level, such as when the capacity drops to 90% or 85% of the initial value. The choice of this starting point depends on the application requirements and safety standards. The selected cycle period refers to the time series data points after the preset starting point. These data points will be used to train the APSO-SVR model test to ensure that the prediction results can reflect the actual degradation trend of the battery. At the same time, the preset period and the selected cycle period are obtained. When the battery capacity prediction value is greater than the battery capacity failure threshold, the selected cycle period is accumulated with the preset period as a new cycle period to obtain the battery capacity prediction value of the new cycle period, and so on, until the battery capacity prediction value of the new cycle period does not exceed the battery capacity failure threshold. The preset cycle is a fixed number of cycles that are advanced during each prediction. If the predicted battery capacity is still above the set failure threshold, we will select the cycle plus the preset cycle as the new cycle, and again use the APSO-SVR model to predict the battery capacity under this new cycle. Whenever the prediction results indicate that the battery capacity remains within the safe range, that is, greater than the failure threshold, we continue to increase the preset cycle and recalculate. In this way, we can gradually approach the actual end of the battery's life until a prediction shows that the battery capacity prediction for a new cycle does not exceed the stated failure threshold.

[0083] First, when using the APSO-SVR model for prediction, the battery capacity degradation data is normalized:

[0084]

[0085] in, is the capacity of the kth cycle, is the initial capacity of the battery. In this embodiment, the experimental battery capacity degradation data are selected from the battery data sets A01 and A02 of Stanford University and the battery data sets cs2_35 and cs2_36 of the University of Maryland. Figure 2 The battery parameter table shown in the figure shows that 95% represents the number of cycles required for the capacity to decay to 95% of the nominal capacity, and the rest are similar.

[0086] Specifically, for the battery data set of Stanford University, three methods, namely standard PSO-SVR, APSO-SVR and VMD-APSO-SVR of this embodiment, are used to compare and analyze the battery RUL prediction performance of two batteries A01 and A02 of Stanford University. RMSE, MAE, MAPE, ERUL and Five indicators are used to verify the effectiveness. Two different prediction starting points are set for the two groups of experiments (40% training 60% test and 70% training 30% test). 40% and 70% of the sample data are used as the data for training the prediction model, and the remaining data are used as test data to predict RUL and capacity. The experiment design in this section is as follows: the sample data and the smoothed data are normalized to 0-1, and e-SVR is selected to build a black box model with the cycle as input and the capacity as output. The loss function is set to 0.001, the kernel function is set to the radial basis kernel function, and the maximum and minimum values ​​of the penalty factor and kernel parameters are set to [ The experimental results are as follows. Figure 3 The A01 battery RUL prediction performance evaluation diagram shown and Figure 4 The A02 battery RUL prediction performance evaluation diagram shown in the figure shows that the VMD-APSO-SVR method has the smallest generalization error and is closest to the real data compared with the other two methods. Therefore, when the amount of data is small in the early stage of degradation, the three methods cannot accurately judge the accelerated degradation process of lithium batteries, but the VMD-APSO-SVR method prediction curve is closest to the real curve, while the PSO-SVR method has the worst fitting effect. Figure 5 A01 battery degradation early and late prediction lithium battery RUL diagram and Figure 6 The A02 battery degradation early and late prediction lithium battery RUL diagram shown and Figure 7 The APSO and PSO fitness curves shown in the figure show that the PSO-SVR, APSO-SVR and VMD-APSO-SVR methods have poor prediction effects in the early stages of degradation, especially in long-term predictions, and cannot predict the accelerated degradation process of lithium batteries. However, the PSO-SVR method has a larger offset in short-term predictions than the other two methods. The overall trend of the predicted value of the APSO-SVR method continues the degradation trend of the training data and has a smaller upward offset from the true value. As the training samples increase from 40% to 70%, the prediction deviation gradually decreases overall, the prediction error begins to decline, all five indicators decrease, and the prediction performance improves. This shows that the more samples, the better the model training, the smaller the generalization error, and the APSO fitness curve has a better effect. Therefore, the lithium-ion battery RUL prediction method based on VMD-APSO-SVR in this embodiment has a small short-term prediction error and high computational efficiency.

[0087] This embodiment provides a battery RUL prediction method based on variable modal decomposition. By obtaining battery capacity degradation data and a battery capacity failure threshold, the battery capacity degradation data is smoothed using variable modal decomposition to obtain multiple modal components. Based on the multiple modal components, key feature data of the battery capacity is obtained. The key feature data of the battery capacity is input into an optimized support vector regression model to obtain a battery capacity prediction value. When the battery capacity prediction value is less than the battery capacity failure threshold, the remaining battery service life is calculated. By using variable modal decomposition to smooth the battery capacity degradation data, key features are extracted and input into the optimized SVR model to predict the battery capacity. When the predicted battery capacity is lower than the failure threshold, the remaining battery service life is calculated, thereby improving the RUL prediction accuracy and robustness.

[0088] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above embodiment 1 can be referred to the above introduction and will not be described in detail later. Figure 8 The battery RUL prediction method based on variable mode decomposition further includes steps S401 to S403 before step S40:

[0089] Step S401: Divide the key characteristic data of battery capacity into a training set and a test set.

[0090] It should be noted that, based on the chronological order or random sampling method, a part of the data is divided out as a training set from all the data, and this part of the data is used to train the model. The training set contains paired data of historical cycle periods and corresponding battery capacities. Through these data, the APSO-SVR model can learn the patterns and laws of battery degradation. At the same time, another part of the data is divided into a test set. The data in the test set does not participate in the training process of the model, but is used to test how the trained model performs on unseen data. Doing so can avoid the occurrence of overfitting, that is, the model only remembers the training data but fails to truly grasp the logical relationship behind it. By comparing the differences between the predicted results of the model on the test set and the actual values, such as the root mean square error (RMSE), the coefficient of determination ( ) and other indicators can be used to quantitatively evaluate the prediction accuracy and robustness of the model.

[0091] Step S402: Acquire training set feature data and establish an initial support vector regression model.

[0092] Step S403 : optimizing the initial support vector regression model according to the feature data of the training set using a particle swarm optimization algorithm to obtain an optimized support vector regression model.

[0093] It should be noted that before obtaining the optimized support vector regression model, its parameters, namely the parameter penalty factor and the kernel parameter, need to be determined.

[0094] Furthermore, before step S403, it includes: obtaining the particle filter in the training set feature data; initializing the algorithm parameters, setting the species measurement threshold; dividing the particle population in the particle filter into multiple subgroups; calculating the fitness value of each particle in each subgroup, and initializing the optimal position of each subgroup individual and the optimal position of the group; updating the fitness value of each subgroup individual, the optimal position of each subgroup individual and the optimal position of the group according to the functional relationship; judging whether the species measurement value exceeds the species measurement threshold, if the particle species measurement value is less than the species measurement threshold, re-planning the particle population; judging whether the maximum number of iterations is reached, if so, outputting the model parameters; after inputting the model parameters into the nonlinear fitting function, obtaining the parameter penalty factor and kernel parameters for constructing the optimized support vector model. Specifically, the algorithm parameters are initialized. Including the total number of particles M, the maximum number of iterations , the maximum value of inertia weight and minimum value , learning factors c1, c2, particle position value , the speed value is , setting a population diversity metric threshold. The particle population in the particle filter is divided into multiple sub-swarms to more efficiently explore the search space. Particles within each sub-swarm interact according to their own position update rules and calculate their own fitness values. The fitness function here measures the quality of a particle's solution to a given problem. For example, in this study, it could be a measure of the error relative to the actual battery capacity. After the initialization phase, each sub-swarm records its individual best position PP(t) and group best position PG(t). These two positions represent the best solutions found so far by a single particle and the entire swarm, respectively. As the iteration process progresses, the speed and position of particles within each sub-swarm are continuously updated according to a predefined functional relationship. Fitness values ​​are recalculated, and the best positions of individuals and groups are adjusted accordingly. This update mechanism aims to guide particles towards better solutions, gradually approaching the global optimal solution. During this process, the particle diversity metric is regularly checked to see if it exceeds a preset threshold. If the particle diversity metric is found to be below the threshold, it indicates insufficient particle diversity, which may lead to premature maturation. In this case, the particle population needs to be replanned, introducing new mutation or crossover operations to increase diversity. When the maximum number of iterations is reached, the optimization process ends, and the output is the optimal model parameters obtained after multiple rounds of iterations—the penalty factor C and the kernel parameter g. These parameters are then input into the nonlinear fitting function to construct the final optimized support vector regression model.

[0095] Furthermore, step S403 also includes: obtaining a particle swarm from the training set data; initializing the positions and velocities of the particles in the swarm and calculating the particle fitness; setting the number of iterations and calculating the historical optimal values ​​of the particles; comparing the fitness of the particles in each iteration to obtain the global optimal value of the particle swarm; updating the particle position and velocity based on the historical optimal value and the global optimal value; determining whether the particle position and velocity meet a preset range; if not, performing boundary condition processing on the particle; and obtaining an optimized support vector regression model if the number of iterations is reached or the error is less than a preset error. Specifically, in the initial stage, each particle is randomly assigned an initial position and velocity. The position represents the coordinates of the particle in the search space and corresponds to a set of SVR parameters; the velocity determines the direction and distance the particle moves in subsequent iterations. Simultaneously, the fitness value of each particle is calculated. This is typically done by using the particle position as an APSO-SVR model parameter to assess the model's fit to the training data, such as minimizing the mean squared error (MSE) or maximizing the cross-validation score. Next, a reasonable number of iterations is set as one of the termination criteria, and the iterative process begins. In each iteration, the fitness of a particle is calculated based on its current position and compared with its previous optimal value. If the new fitness is better, the previous optimal value is updated. Simultaneously, the previous optimal values ​​of all particles are used to update the global optimal value—the best solution found by the entire swarm. The global optimal value reflects the best parameter combination explored so far by all particles. As iterations proceed, the position and velocity of each particle are adjusted according to a specific velocity update formula based on the previous and global optimal values. This process simulates information sharing and learning between particles, enabling particles to move toward more optimal solutions. For example, the new velocity of a particle can be a weighted sum of its current velocity, its attraction toward the previous optimal value, and its attraction toward the global optimal value. To ensure that particles remain within a reasonable range, it is necessary to determine whether their positions and velocities meet a preset range. If not, boundary conditions are applied to the particles, such as bounce, truncation, or re-randomization, to prevent them from escaping the effective search space, thereby ensuring the algorithm's effectiveness and stability. When the set number of iterations is reached or the error is less than the preset error, the optimization process ends and the optimized support vector regression model is output.

[0096] In this embodiment, the key characteristic data of battery capacity are divided into a training set and a test set, and the initial support vector regression model is established using the training set and optimized using the particle swarm optimization algorithm to improve the prediction accuracy and model robustness.

[0097] Based on the first embodiment of the present application, in the third embodiment of the present application, the same or similar contents as those in the first embodiment can be referred to the above introduction and will not be described in detail later. Figure 9The battery RUL prediction method based on variable mode decomposition step S20 further includes steps S201 to S203:

[0098] Step S201: Obtain modal decomposition parameters.

[0099] It's important to note that modal decomposition parameters are key settings in the Variational Mode Decomposition (VMD) algorithm, used to decompose complex signals, such as lithium-ion battery capacity degradation data, into a series of intrinsic mode functions (IMFs) with varying center frequencies. To obtain these parameters, the first step is to determine the number of modal components in the sequence to be decomposed—that is, the desired number of IMF components. This choice depends on the complexity of the original signal and the purpose of the analysis. For battery capacity prediction, different numbers of modal components are typically tried to find the set of IMFs that best reflects battery aging characteristics.

[0100] Another important parameter is the weight, which controls the weight of the bandwidth penalty term and affects the smoothness of each IMF component. Larger weight values ​​result in a narrower bandwidth, making the decomposed IMF smoother but potentially losing some detail. Conversely, smaller weight values ​​allow for a wider bandwidth, preserving more high-frequency components but potentially introducing noise. Therefore, in practical applications, this value needs to be adjusted based on specific circumstances to achieve the optimal balance.

[0101] Furthermore, the time step is a crucial parameter, determining the accuracy of the instantaneous phase estimate during the iterations. Properly setting the time step can accelerate convergence without sacrificing decomposition quality. Finally, the initialization noise tolerance serves as a termination criterion; the algorithm is considered converged when the change between two consecutive iterations is less than this threshold.

[0102] Step S202: Obtain the number of modal components and the center frequencies of the modal components according to the modal decomposition parameters.

[0103] It's important to note that after setting the key VMD parameters, the algorithm iteratively searches for the optimal solution, ultimately outputting multiple IMFs and their corresponding center frequencies. These IMFs each represent information from a different frequency band in the original signal, encompassing a range of fluctuations from high to low frequencies. Specifically, when the appropriate number of modal components is selected, VMD can effectively isolate the various levels of variation associated with battery aging. For example, in lithium-ion battery research, some IMFs may exhibit rapid variations, corresponding to short-term charge-discharge cycling effects, while other, lower-frequency IMFs reflect long-term aging patterns. Each IMF has a unique center frequency, which helps distinguish different types of degradation behavior and can reveal the physical mechanisms underlying complex data. By analyzing the resulting modal components and their center frequencies, we can better understand the temporal evolution of battery performance. For example, if the center frequency of a particular IMF approaches or falls below a critical value, it may be due to a slowdown in the battery's internal chemical reaction rate, indicating that the battery is about to enter a stage of accelerated degradation.

[0104] Step S203 : Iteratively decompose the battery capacity degradation data based on the number of modal components and the center frequency to extract multiple modal components.

[0105] It should be noted that VMD decomposes the original battery capacity degradation time series into multiple intrinsic mode functions (IMFs), each of which represents the fluctuation characteristics within a specific frequency band. This decomposition is not a one-time process but is completed gradually through an iterative optimization process.

[0106] Furthermore, in each iteration, the VMD algorithm attempts to minimize the sum of the bandwidths of all IMFs while ensuring that the sum of these IMFs can reconstruct the original signal. For each IMF, its Hilbert transform relative to the instantaneous phase is calculated and modulated to baseband to obtain spectral information. Next, the bandwidth of each IMF is estimated and its center frequency is updated. As the number of iterations increases, the individual IMFs gradually converge to the optimal solution, ultimately obtaining a set of smooth and independent IMF components that collectively reflect the distinct characteristics of battery capacity variation over time. From these extracted IMFs, it is possible to further filter those that best characterize battery aging behavior, such as low-frequency IMFs that exhibit slow, long-term variations. This is because high-frequency components often contain more noise or short-term disturbances, while low-frequency IMFs more closely reflect the actual battery aging pattern. Furthermore, the Pearson correlation coefficient can be used to measure the correlation between each IMF and the original signal. IMFs with strong or extremely strong correlations are selected as the basis for reconstructing the lithium battery capacity data, thereby improving the accuracy of subsequent analysis.

[0107] This embodiment obtains modal decomposition parameters, determines the number of modal components and center frequency, and based on this, iteratively decomposes the battery capacity degradation data to extract multiple modal components, effectively remove noise, and improve the accuracy and robustness of the prediction model.

[0108] This application also provides a battery RUL prediction device based on variable mode decomposition, please refer to Figure 10 , the device comprises:

[0109] The acquisition module 10 is used to acquire battery capacity degradation data and a battery capacity failure threshold.

[0110] The processing module 20 is used to smooth the battery capacity degradation data by using variable mode decomposition to obtain multiple modal components.

[0111] The extraction module 30 is used to obtain key characteristic data of battery capacity according to multiple modal components.

[0112] The calculation module 40 is used to obtain a predicted value of the battery capacity by inputting key characteristic data of the battery capacity into the optimized support vector regression model.

[0113] The result module 50 is used to calculate the remaining service life of the battery when the battery capacity prediction value is less than the battery capacity failure threshold.

[0114] The battery RUL prediction device based on variable modal decomposition provided in this application adopts the battery RUL prediction method based on variable modal decomposition in the above embodiment, which can solve the technical problem of how to improve the accuracy of battery RUL prediction. Compared with the prior art, the beneficial effects of the battery RUL prediction device based on variable modal decomposition provided in this application are the same as the beneficial effects of the battery RUL prediction method based on variable modal decomposition provided in the above embodiment, and the other technical features of the battery RUL prediction device based on variable modal decomposition are the same as the features disclosed in the above embodiment method, and will not be repeated here.

[0115] In one embodiment, the calculation module 40 is further used to divide the key characteristic data of the battery capacity into a training set and a test set; obtain the characteristic data of the training set and establish an initial support vector regression model; optimize the initial support vector regression model according to the particle swarm optimization algorithm based on the characteristic data of the training set to obtain an optimized support vector regression model.

[0116] In one embodiment, the calculation module 40 is also used to obtain the particle filter in the training set feature data; initialize the algorithm parameters and set the species measurement threshold; divide the particle population in the particle filter into multiple subgroups; calculate the fitness value of each particle in each subgroup, and initialize the optimal position of each subgroup individual and the optimal position of the group; update the fitness value of each subgroup individual, the optimal position of each subgroup individual and the optimal position of the group according to the functional relationship; determine whether the species measurement value exceeds the species measurement threshold, if the particle species measurement value is less than the species measurement threshold, re-plan the particle population; determine whether the maximum number of iterations is reached, if so, output the model parameters; after inputting the model parameters into the nonlinear fitting function, obtain the parameter penalty factor and kernel parameters for constructing the optimized support vector model.

[0117] In one embodiment, the calculation module 40 is further used to obtain a particle swarm from the training set data information; initialize the position and velocity of particles in the particle swarm and calculate the fitness of the particles; set the number of iterations and calculate the historical optimal value of the particles; obtain the global optimal value of the particle swarm by comparing the fitness of the particles in each iteration; update the position and velocity of the particles based on the historical optimal value and the global optimal value; determine whether the position and velocity of the particles meet a preset range, and if not, perform boundary condition processing on the particles; if the number of iterations is reached or the error is less than the preset error, an optimized support vector regression model is obtained.

[0118] In one embodiment, the calculation module 40 is further used to obtain the preset starting point cycle period and the selected cycle period; when the selected cycle period exceeds the cycle period of the preset starting point, the selected cycle period is substituted into the optimized support vector regression model to obtain the battery capacity prediction value.

[0119] In one embodiment, the processing module 20 is further used to obtain modal decomposition parameters; obtain the number of modal components and the center frequency of the modal components based on the modal decomposition parameters; iteratively decompose the battery capacity degradation data based on the number of modal components and the center frequency to extract multiple modal components.

[0120] In one embodiment, the result module 50 is further used to obtain a preset period and select a cycle period; when the battery capacity prediction value is greater than the battery capacity failure threshold, the selected cycle period plus the preset period is accumulated as a new cycle period to obtain the battery capacity prediction value of the new cycle period, and so on, until the battery capacity prediction value of the new cycle period does not exceed the battery capacity failure threshold.

[0121] The present application provides a battery RUL prediction device based on variable modal decomposition, and the battery RUL prediction device based on variable modal decomposition includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the battery RUL prediction method based on variable modal decomposition in the above-mentioned embodiment one.

[0122] Reference below Figure 11 , which shows a schematic structural diagram of a battery RUL prediction device based on variable modal decomposition suitable for implementing an embodiment of the present application. The battery RUL prediction device based on variable modal decomposition in the embodiment of the present application may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), in-vehicle terminals (such as in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 11 The battery RUL prediction device based on variable modal decomposition shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0123] like Figure 11As shown, the battery RUL prediction device based on variable modal decomposition may include a processing device 1001 (such as a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM: Read Only Memory) 1002 or the program loaded from the storage device 1003 to the random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of the battery RUL prediction device based on variable modal decomposition are also stored. The processing device 1001, ROM1002 and RAM1004 are connected to each other via a bus 1005. The input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the battery RUL prediction device based on variable mode decomposition to communicate wirelessly or wired with other devices to exchange data. Although the figure shows a battery RUL prediction device based on variable mode decomposition with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems may be implemented or have instead.

[0124] In particular, according to the embodiments disclosed in the present application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments disclosed in the present application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program comprising program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0125] The battery RUL prediction device based on variable modal decomposition provided in this application adopts the battery RUL prediction method based on variable modal decomposition in the above embodiment, which can solve the technical problem of how to improve the accuracy of battery RUL prediction. Compared with the prior art, the beneficial effects of the battery RUL prediction device based on variable modal decomposition provided in this application are the same as the beneficial effects of the battery RUL prediction method based on variable modal decomposition provided in the above embodiment, and the other technical features of the battery RUL prediction device based on variable modal decomposition are the same as the features disclosed in the method of the previous embodiment, and will not be repeated here.

[0126] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0127] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0128] The present application provides a computer-readable medium having computer-readable program instructions (ie, a computer program) stored thereon, and the computer-readable program instructions are used to execute the battery RUL prediction method based on variable modal decomposition in the above embodiment.

[0129] The computer-readable medium provided in this application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination thereof. More specific examples of computer-readable media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this embodiment, the computer-readable medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0130] The computer-readable medium may be included in the battery RUL prediction device based on variable mode decomposition; or may exist independently without being assembled into the battery RUL prediction device based on variable mode decomposition.

[0131] The computer-readable medium carries one or more programs. When executed by a battery RUL prediction device based on variable modal decomposition, the one or more programs enable the battery RUL prediction device based on variable modal decomposition to write computer program code for performing the operations of the present application in one or more programming languages, or a combination thereof. The programming languages ​​include object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as "C" or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0132] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a module, program segment or a part of code, and the module, program segment or a part of code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented by a dedicated hardware-based system that performs the specified function or operation, or can be implemented by a combination of dedicated hardware and computer instructions.

[0133] The modules described in the embodiments of the present application may be implemented in software or hardware, wherein the name of a module does not necessarily limit the unit itself.

[0134] The computer-readable medium provided in this application is a computer-readable medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described battery RUL prediction method based on variable modal decomposition. This computer-readable medium can solve the technical problem of how to improve the accuracy of battery RUL prediction. Compared with the prior art, the beneficial effects of the computer-readable medium provided in this application are the same as those of the battery RUL prediction method based on variable modal decomposition provided in the above-described embodiment, and are not further described here.

[0135] The present application also provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of the battery RUL prediction method based on variable mode decomposition as described above.

[0136] The computer program product provided in this application can solve the technical problem of how to improve the accuracy of battery RUL prediction. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the battery RUL prediction method based on variable mode decomposition provided in the above embodiment, and will not be repeated here.

[0137] The above description is only part of the embodiments of the present application and does not limit the patent scope of the present application. All equivalent structural transformations made by using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect application in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A battery RUL prediction method based on variable mode decomposition, characterized in that: The method comprises: Obtain battery capacity degradation data and battery capacity failure threshold; Smoothing the battery capacity degradation data using variable mode decomposition to obtain multiple modal components; Obtaining key characteristic data of battery capacity according to the plurality of modal components; Obtaining a predicted value of battery capacity by inputting the key characteristic data of battery capacity into an optimized support vector regression model; When the battery capacity prediction value is less than the battery capacity failure threshold, calculating the remaining battery service life; Before the step of inputting the key characteristic data of the battery capacity into the optimized support vector regression model to obtain the predicted value of the battery capacity, the method includes: Dividing the battery capacity key feature data into a training set and a test set; Obtain training set feature data and establish an initial support vector regression model; Optimizing the initial support vector regression model according to the training set feature data using a particle swarm optimization algorithm to obtain an optimized support vector regression model; The step of obtaining a predicted value of battery capacity by inputting the key characteristic data of battery capacity into an optimized support vector regression model comprises: Get the preset starting point cycle period and the selected cycle period; When the selected cycle period exceeds the cycle period of the preset starting point, substituting the selected cycle period into the optimized support vector regression model to obtain a predicted value of battery capacity; Before the step of optimizing the initial support vector regression model according to the training set feature data using a particle swarm optimization algorithm to obtain an optimized support vector regression model, the method includes: Obtain particle filtering in the training set feature data; Initialize algorithm parameters and set category measurement thresholds; Dividing the particle population in the particle filter into multiple subgroups; Calculate the fitness value of each particle in each subgroup, and initialize the best position of each subgroup individual and the best position of the group; Update the fitness value of each subgroup individual, the best position of each subgroup individual and the best position of the group according to the functional relationship; Determine whether the type metric value exceeds the type metric threshold. If the particle type metric value is less than the type metric threshold, re-plan the particle population. Determine whether the maximum number of iterations has been reached. If so, output the model parameters. After inputting the model parameters into the nonlinear fitting function, the parameter penalty factor and kernel parameter for constructing the optimized support vector model are obtained; The step of optimizing the initial support vector regression model according to the training set feature data using a particle swarm optimization algorithm to obtain an optimized support vector regression model includes: Obtaining a particle swarm from the training set data information; Initializing the positions and velocities of particles in the particle swarm and calculating the fitness of the particles; Set the number of iterations and calculate the historical optimal value of the particle; By comparing the fitness of the particles in each iteration, a global optimal value of the particle swarm is obtained; updating the position and velocity of the particle according to the historical optimal value and the global optimal value; Determine whether the position and velocity of the particle meet a preset range, and if not, perform boundary condition processing on the particle; If the number of iterations is reached or the error is less than the preset error, the optimized support vector regression model is obtained; The step of smoothing the battery capacity degradation data using variable mode decomposition to obtain multiple modal components includes: Get modal decomposition parameters; According to the modal decomposition parameters, the number of modal components and the center frequencies of the modal components are obtained; Iteratively decomposing the battery capacity degradation data based on the number of modal components and the center frequency to extract multiple modal components; The step of calculating the remaining battery service life when the battery capacity prediction value is less than the battery capacity failure threshold further includes: Get preset period and select cycle period; When the battery capacity prediction value is greater than the battery capacity failure threshold, the selected cycle period is added to the preset cycle period as a new cycle period to obtain the battery capacity prediction value of the new cycle period, and so on, until the battery capacity prediction value of the new cycle period does not exceed the battery capacity failure threshold.

2. A battery RUL prediction device based on variable mode decomposition, characterized in that: The device comprises: An acquisition module, used to obtain battery capacity degradation data and battery capacity failure threshold; a processing module, configured to smooth the battery capacity degradation data using variable modal decomposition to obtain a plurality of modal components; The processing module is further configured to obtain modal decomposition parameters; obtain the number of modal components and the center frequencies of the modal components according to the modal decomposition parameters; and iteratively decompose the battery capacity degradation data based on the number of modal components and the center frequencies to extract multiple modal components; an extraction module, configured to obtain key characteristic data of battery capacity according to the plurality of modal components; a calculation module, configured to obtain a predicted value of battery capacity by inputting the key characteristic data of battery capacity into an optimized support vector regression model; The calculation module is further used to divide the key characteristic data of the battery capacity into a training set and a test set; Obtain training set feature data and establish an initial support vector regression model; Optimizing the initial support vector regression model according to the training set feature data using a particle swarm optimization algorithm to obtain an optimized support vector regression model; The calculation module is also used to obtain the preset starting point cycle period and select the cycle period; When the selected cycle period exceeds the cycle period of the preset starting point, substituting the selected cycle period into the optimized support vector regression model to obtain a predicted value of battery capacity; The calculation module is also used to obtain the particle filter in the training set feature data; initialize the algorithm parameters and set the species measurement threshold; divide the particle population in the particle filter into multiple subgroups; calculate the fitness value of each particle in each subgroup, and initialize the optimal position of each subgroup individual and the optimal position of the group; update the fitness value of each subgroup individual, the optimal position of each subgroup individual and the optimal position of the group according to the functional relationship; determine whether the species measurement value exceeds the species measurement threshold, if the particle species measurement value is less than the species measurement threshold, re-plan the particle population; determine whether the maximum number of iterations has been reached, and if so, output the model parameters; after inputting the model parameters into the nonlinear fitting function, obtain the parameter penalty factor and kernel parameters for constructing the optimized support vector model; The calculation module is further used to obtain a particle swarm from the training set data information; initialize the position and velocity of particles in the particle swarm and calculate the fitness of the particles; set the number of iterations and calculate the historical optimal value of the particles; obtain the global optimal value of the particle swarm by comparing the fitness of the particles in each iteration; update the position and velocity of the particles according to the historical optimal value and the global optimal value; determine whether the position and velocity of the particles meet a preset range, and if not, perform boundary condition processing on the particles; if the number of iterations is reached or the error is less than the preset error, obtain the optimized support vector regression model; A result module is used to calculate the remaining service life of the battery when the battery capacity prediction value is less than the battery capacity failure threshold; The result module is also used to obtain a preset period and a selected cycle period; when the battery capacity prediction value is greater than the battery capacity failure threshold, the selected cycle period is added to the preset period as a new cycle period to obtain the battery capacity prediction value of the new cycle period, and so on, until the battery capacity prediction value of the new cycle period does not exceed the battery capacity failure threshold.

3. A battery RUL prediction device based on variable mode decomposition, characterized in that: The device includes: a memory, a processor, and a battery RUL prediction program based on variable modal decomposition stored in the memory and running on the processor, wherein the battery RUL prediction program based on variable modal decomposition is configured to implement the steps of the battery RUL prediction method based on variable modal decomposition as claimed in claim 1.

4. A storage medium, characterized in that The storage medium stores a battery RUL prediction program based on variable modal decomposition. When the battery RUL prediction program based on variable modal decomposition is executed by the processor, the steps of the battery RUL prediction method based on variable modal decomposition as claimed in claim 1 are implemented.

Images