Lithium ion battery life prediction method and system based on variational mode decomposition

Abstract

The invention provides a lithium ion battery life prediction method and system based on variational mode decomposition, and the method comprises the steps: obtaining a lithium ion battery charge-discharge cycle data set, obtaining a key health factor based on the data set, and carrying out the normalization processing of the key health factor; decomposing the normalized key health factor into a plurality of IMF components through a VMD method; and constructing a corresponding gating circulation unit neural network GRU model for each IMF component, obtaining a predicted value of each IMF component by using the corresponding GRU model, and predicting the life end point of the lithium ion battery based on the predicted value of each IMF component. According to the method, original complex signals are converted into physically interpretable subsequences to be input through the preprocessing process of the VMD, a structured feature space is provided for the GRU network, the coupling method gives full play to the dual advantages of signal analysis and neural network modeling, and the robustness and prediction accuracy of a prediction model in a strong noise environment are remarkably enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of battery health monitoring, and in particular to a method and system for predicting the life of a lithium-ion battery based on variational mode decomposition. Background Art

[0002] Lithium-ion batteries will gradually degrade during the daily charge and discharge cycle, resulting in problems such as increased internal resistance, which will affect the safety and reliability of the energy storage power station operation procedures.

[0003] Existing methods for predicting the lifespan of lithium-ion batteries mainly fall into two categories: model-based methods and data-driven methods. Model-based methods are further divided into mechanism-based models and empirical models. Mechanism-based models are based on the complex internal reactions of lithium-ion batteries. However, due to the difficulty in accurately predicting actual conditions and the large number of influencing factors, establishing an accurate mechanism-based model is difficult. Empirical models use formulas fitted according to battery operating data, are not universal, and have poor scalability. Therefore, data-driven methods, due to their advantage of not requiring analysis of the internal mechanisms of the battery, have strong practicality and have been widely used in SOH (State of Health) estimation in recent years. However, how to provide an effective way to predict the performance of lithium-ion batteries in future operations based on data from their daily operations, thereby achieving better management of power plants, extending their service life, and increasing their safety, remains an important challenge that needs to be overcome in the future. Summary of the Invention

[0004] The purpose of the present invention is to provide a lithium-ion battery life prediction method and system based on variational mode decomposition, aiming to solve the above-mentioned problems in the prior art.

[0005] An embodiment of the present invention provides a lithium-ion battery life prediction method based on variational mode decomposition, comprising: Acquire a lithium-ion battery charge and discharge cycle dataset, obtain key health factors based on the dataset, and normalize the key health factors; The normalized key health factors are decomposed into several intrinsic mode function (IMF) components using the VMD method. A corresponding gated recurrent unit neural network (GRU) model is constructed for each IMF component, and the corresponding GRU model is used to obtain the predicted value of each IMF component. The end of life of the lithium-ion battery is predicted based on the predicted value of each IMF component.

[0006] An embodiment of the present invention provides a lithium-ion battery life prediction system based on variational mode decomposition, comprising: A data module is used to obtain a data set of lithium-ion battery charge and discharge cycles, obtain key health factors based on the data set, and normalize the key health factors; A decomposition module is used to decompose the normalized key health factors into several intrinsic mode function (IMF) components using the VMD method; The prediction module is used to construct a corresponding gated recurrent unit neural network GRU model for each IMF component, and use the corresponding GRU model to obtain the predicted value of each IMF component, and predict the end of the life of the lithium-ion battery based on the predicted value of each IMF component.

[0007] An embodiment of the present invention also provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, the steps of the above-mentioned lithium-ion battery life prediction method based on variational mode decomposition are implemented.

[0008] An embodiment of the present invention further provides a computer-readable storage medium, on which a program for implementing information transmission is stored. When the program is executed by a processor, the steps of the above-mentioned lithium-ion battery life prediction method based on variational mode decomposition are implemented.

[0009] The use of the embodiment of the present invention can include the following beneficial effects: the embodiment of the present invention converts the original complex signal into a physically interpretable subsequence input through the VMD preprocessing process, providing a structured feature space for the GRU network. This coupling method gives full play to the dual advantages of signal analysis and neural network modeling, enabling the entire prediction system to simultaneously identify slow-changing trends and rapid fluctuation characteristics in the degradation process, significantly enhancing the robustness and prediction accuracy of the prediction model in a strong noise environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate one or more embodiments of this specification 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, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0011] Figure 1 This is a flow chart of a lithium-ion battery life prediction method based on variational mode decomposition according to an embodiment of the present invention; Figure 2 1. This is a schematic diagram of the complete steps of the lithium-ion battery life prediction method according to an embodiment of the present invention; Figure 34 is a schematic diagram of a lithium-ion battery life prediction system based on variational mode decomposition according to an embodiment of the present invention. DETAILED DESCRIPTION

[0012] In order to enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the technical solutions in one or more embodiments of this specification will be clearly and completely described below in conjunction with the drawings in one or more embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of this document.

[0013] Method Example According to an embodiment of the present invention, a lithium-ion battery life prediction method based on variational mode decomposition is provided. Figure 1 FIG. 1 is a flow chart of a lithium-ion battery life prediction method based on variational mode decomposition according to an embodiment of the present invention. Figure 1 As shown, the lithium-ion battery life prediction method based on variational mode decomposition according to an embodiment of the present invention specifically includes: Step S101, obtaining a lithium-ion battery charge and discharge cycle dataset, obtaining key health factors based on the dataset, and normalizing the key health factors, specifically includes: Extracting a discharge residual capacity sequence for each cycle based on the data set, and calculating a battery internal resistance sequence for each cycle based on the charge and discharge curves in the data set, verifying a strong correlation between the discharge residual capacity sequence and the battery internal resistance sequence according to a Pearson correlation coefficient, using the verified battery internal resistance sequence as a key health factor, normalizing the key health factor, and converting the normalized key health factor into a lithium-ion battery state of health (SOH) using Formula 1; Formula 1; in, Indicates the health status of the battery at the tth cycle, represents the true normalized internal resistance of the t-th cycle.

[0014] Step S102: Decomposing the normalized key health factors into several intrinsic mode function (IMF) components using the VMD method, specifically including: According to the decomposition requirements, the normalized key health factors are decomposed into several intrinsic mode function (IMF) components using the VMD method; Among them, each IMF component is distributed from high to low according to frequency.

[0015] Step S103, constructing a corresponding gated recurrent unit neural network GRU model for each IMF component, and using the corresponding GRU model to obtain a predicted value for each IMF component, and predicting the end of life of the lithium-ion battery based on the predicted value of each IMF component, specifically including: Divide the data set for each IMF component to generate respective training sets and test sets. Use the training sets of each IMF component to independently train a GRU model to obtain a trained GRU model corresponding to each IMF component. Use the test sets of each IMF component to use the trained GRU model corresponding to each IMF component to obtain a predicted value for each IMF component. Superimpose the predicted values ​​of each IMF component to reconstruct a normalized internal resistance series predicted value. Denormalize the normalized internal resistance series predicted value using Formula 2 to obtain a battery state of health (SOH) predicted value. Formula 2; in, represents the battery health status predicted at the tth cycle, represents the normalized internal resistance sequence predicted at the tth cycle; Performing an end-of-life determination on the lithium-ion battery based on the predicted SOH value; if the predicted SOH value is less than or equal to a preset threshold, it indicates that the lithium-ion battery has reached the end of life, and marking the current cycle number as the end-of-life cycle number; If the SOH prediction value is greater than a preset threshold, it indicates that the current lithium-ion battery is in a healthy state.

[0016] The above technical solution of the embodiment of the present invention is described in detail below in conjunction with the specific situation of the lithium-ion battery life prediction method based on variational mode decomposition according to the embodiment of the present invention.

[0017] Variational mode decomposition (VMD) technology decomposes complex battery capacity degradation sequences into several intrinsic mode function components through an adaptive variational framework. This decomposition process can effectively separate physical components such as long-term attenuation trends, medium-term cyclic fluctuations, and short-term random noise in the signal, providing feature inputs with clear physical meaning for subsequent prediction models. The gated recurrent unit (GRU) neural network, as the core engine of sequence modeling, efficiently learns time series features through an innovative gating mechanism. Its unique update gate and reset gate structure can dynamically adjust the weight distribution of historical states and current inputs, significantly reducing model complexity while ensuring long-term memory capabilities. This architecture demonstrates excellent nonlinear sequence modeling capabilities and is particularly suitable for prediction scenarios such as battery degradation, which have strong nonlinear and time-dependent characteristics.

[0018] Therefore, the embodiment of the present invention proposes a lithium battery life prediction method based on VMD-GRU, the steps comprising: Step 1: Data acquisition and preprocessing A. Input data: Obtain a publicly available lithium-ion battery charge-discharge cycle dataset (such as the NASA PCoC dataset) and extract the remaining discharge capacity sequence C(t) for each cycle (where t is the number of cycles).

[0019] B. Internal resistance calculation: Based on the charge and discharge curve (voltage, current, time), calculate the battery internal resistance R(t) (key health factor) for each cycle using Ohm's law or electrochemical impedance spectroscopy (EIS).

[0020] C. Correlation Verification: Calculate the Pearson correlation coefficient between internal resistance R(t) and remaining capacity C(t), confirming a strong correlation between the two (|r|>0.8), verifying the validity of internal resistance as a health factor.

[0021] Step 2: Health factor normalization Normalization formula: (1); Mapping the internal resistance to the interval [0, 1] directly represents the state of health (SOH): (2).

[0022] Step 3: Variational Mode Decomposition (VMD) A. Decomposition goal: Normalize the internal resistance sequence Decomposed into 5 intrinsic mode functions (IMFs); (3); B. Parameter setting: VMD mode number k=5, penalty factor =2000, the convergence tolerance is ; C. Decomposition effect: Each IMF component must satisfy the frequency distribution from high to low, with high-frequency components capturing short-term fluctuations (such as temperature effects) and low-frequency components representing long-term aging trends.

[0023] Step 4: Data division and GRU modeling A. Dataset Partitioning: For each IMF component, divide the training set and test set into a ratio of 7:3 (the time sequence cannot be disrupted).

[0024] B. GRU model construction: Build an independent GRU model for each IMF component, with the following structure: Input layer: sliding window length T=10 (historical 10 cycle data); Hidden layer: 2 layers of GRU units (64 neurons per layer); Output layer: single neuron (predicts the IMF value of the next cycle).

[0025] C. Hyperparameter optimization: Use Bayesian optimization to determine the learning rate (0.001–0.01) and dropout rate (0.2–0.5).

[0026] Step 5: Component prediction and reconstruction A. Single component prediction: Use the trained GRU model to predict the future values ​​of each IMF in the test set .

[0027] B. Reconstruct the internal resistance sequence: Adding up all the IMF forecasts: (4); SOH denormalization: (5).

[0028] Step 6: End of life prediction Termination threshold determination: when When , mark the current cycle number t as the predicted life end.

[0029] The following is a complete step supplement to the lithium-ion battery life prediction method proposed in an embodiment of the present invention and a logical summary of the technical solution: like Figure 2 As shown, the embodiment of the present invention specifically includes the following steps: 1. Data Acquisition: We obtained NASA's public lithium-ion battery dataset and processed it to obtain battery health factors such as voltage, current, and internal resistance. We calculated the correlation coefficient to confirm that the discharge internal resistance data series is strongly correlated with battery health. 2. Data processing: Based on the VMD method, the battery internal resistance series is decomposed into five less complex and relatively stable IMF components: , the meanings of the five components are as follows: IMF1-2: high-frequency components (sensor noise); IMF3-4: intermediate frequency component (affected by temperature fluctuation); IMF5: low-frequency component (battery aging trend); 3. Model training: First, the five components after decomposition are divided into training sets and test sets according to a certain ratio. Then, a GRU model is established for each component. The hyperparameters of the GRU network are obtained by training the model. The trained model is used to predict the future value of the battery internal resistance to obtain the predicted value of each IMF component. The predicted values ​​of each component predicted by GRU are accumulated to obtain the predicted value of the residual internal resistance of the lithium-ion battery after discharge. After calculation, the battery's SOH can be obtained.

[0030] The key points of technical logic are as follows: 1. Relationship between internal resistance and capacity Increased internal resistance is the direct physical manifestation of battery aging, while capacity decay is a consequence. The present invention uses internal resistance as the core health factor and, through Pearson correlation verification, avoids interference phenomena such as capacity regeneration.

[0031] 2. Necessity of VMD decomposition Solve the non-stationary problem of internal resistance series (confusion of trend, noise, and periodic fluctuation).

[0032] Separating aging factors at different time scales enables GRU to learn the laws of each component in a targeted manner.

[0033] 3. Lifespan prediction process Internal resistance calculation → normalization → VMD decomposition → component prediction → superposition reconstruction → SOH mapping → end-of-life determination, forming a closed-loop logic chain, and ultimately outputting a quantifiable number of cycle life.

[0034] System Example According to an embodiment of the present invention, a lithium-ion battery life prediction system based on variational mode decomposition is provided. Figure 3 Schematic diagram of a lithium-ion battery life prediction system based on variational mode decomposition according to an embodiment of the present invention. Figure 3 As shown, the lithium-ion battery life prediction system based on variational mode decomposition according to an embodiment of the present invention specifically includes: The data module 30 is used to obtain a lithium-ion battery charge and discharge cycle data set, obtain key health factors based on the data set, and normalize the key health factors, specifically for: Extracting a discharge residual capacity sequence for each cycle based on the data set, and calculating a battery internal resistance sequence for each cycle based on the charge and discharge curves in the data set, verifying a strong correlation between the discharge residual capacity sequence and the battery internal resistance sequence according to a Pearson correlation coefficient, using the verified battery internal resistance sequence as a key health factor, normalizing the key health factor, and converting the normalized key health factor into a lithium-ion battery state of health (SOH) using Formula 1; Formula 1; in, Indicates the health status of the battery at the tth cycle, represents the true normalized internal resistance of the t-th cycle.

[0035] The decomposition module 32 is used to decompose the normalized key health factors into a number of intrinsic mode function (IMF) components using the VMD method, specifically for: According to the decomposition requirements, the normalized key health factors are decomposed into several intrinsic mode function (IMF) components using the VMD method; Among them, each IMF component is distributed from high to low according to frequency.

[0036] The prediction module 34 is used to construct a corresponding gated recurrent unit neural network GRU model for each IMF component, and use the corresponding GRU model to obtain a predicted value for each IMF component, and predict the end of life of the lithium-ion battery based on the predicted value of each IMF component, specifically for: Divide the data set for each IMF component to generate respective training sets and test sets. Use the training sets of each IMF component to independently train a GRU model to obtain a trained GRU model corresponding to each IMF component. Use the test sets of each IMF component to use the trained GRU model corresponding to each IMF component to obtain a predicted value for each IMF component. Superimpose the predicted values ​​of each IMF component to reconstruct a normalized internal resistance series predicted value. Denormalize the normalized internal resistance series predicted value using Formula 2 to obtain a battery state of health (SOH) predicted value. Formula 2; in, represents the battery health status predicted at the tth cycle, represents the normalized internal resistance sequence predicted at the tth cycle; Performing an end-of-life determination on the lithium-ion battery based on the predicted SOH value; if the predicted SOH value is less than or equal to a preset threshold, it indicates that the lithium-ion battery has reached the end of life, and marking the current cycle number as the end-of-life cycle number; If the SOH prediction value is greater than a preset threshold, it indicates that the current lithium-ion battery is in a healthy state.

[0037] The embodiment of the present invention is a system embodiment corresponding to the above-mentioned method embodiment. The specific operations of each module can be understood by referring to the description of the method embodiment, which will not be repeated here.

[0038] In summary, in the collaborative working mechanism of the embodiment of the present invention, the VMD preprocessing process converts the original complex signal into a physically interpretable subsequence input, providing a structured feature space for the GRU network. This coupling method fully leverages the dual advantages of signal analysis and neural network modeling, enabling the entire prediction system to simultaneously identify slow-changing trends and rapid fluctuation characteristics in the degradation process. This technical path significantly enhances the robustness and prediction accuracy of the prediction model in a strong noise environment, providing a reliable technical guarantee for battery health status assessment.

[0039] Device Example 1 An embodiment of the present invention provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program implements the steps described in the method embodiment when executed by the processor.

[0040] Device Example 2 An embodiment of the present invention provides a computer-readable storage medium, on which a program for implementing information transmission is stored. When the program is executed by a processor, the steps described in the method embodiment are implemented.

[0041] The computer-readable storage medium in this embodiment includes, but is not limited to, ROM, RAM, magnetic disk, or optical disk.

[0042] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A lithium-ion battery life prediction method based on variational mode decomposition (VMD), characterized by include: Acquire a lithium-ion battery charge and discharge cycle dataset, obtain key health factors based on the dataset, and normalize the key health factors; The normalized key health factors are decomposed into several intrinsic mode function (IMF) components using the VMD method. A corresponding gated recurrent unit neural network (GRU) model is constructed for each IMF component, and the corresponding GRU model is used to obtain the predicted value of each IMF component. The end of life of the lithium-ion battery is predicted based on the predicted value of each IMF component.

2. The method according to claim 1, characterized in that Obtaining key health factors based on the data set and normalizing the key health factors specifically include: Extracting a discharge residual capacity sequence for each cycle based on the data set, and calculating a battery internal resistance sequence for each cycle based on the charge and discharge curves in the data set, verifying a strong correlation between the discharge residual capacity sequence and the battery internal resistance sequence according to a Pearson correlation coefficient, using the verified battery internal resistance sequence as a key health factor, normalizing the key health factor, and converting the normalized key health factor into a lithium-ion battery state of health (SOH) using Formula 1; Formula 1: in, Indicates the health status of the battery at the tth cycle, represents the true normalized internal resistance of the t-th cycle.

3. The method according to claim 1, characterized in that The normalized key health factors are decomposed into several intrinsic mode function (IMF) components using the VMD method, including: According to the decomposition requirements, the normalized key health factors are decomposed into several intrinsic mode function (IMF) components using the VMD method; Among them, each IMF component is distributed from high to low according to frequency.

4. The method according to claim 1, wherein A corresponding gated recurrent unit neural network GRU model is constructed for each IMF component, and the corresponding GRU model is used to obtain the predicted value of each IMF component. The end of life of the lithium-ion battery is predicted based on the predicted value of each IMF component. Specifically, the following steps are involved: Divide the data set for each IMF component to generate respective training sets and test sets. Use the training sets of each IMF component to independently train a GRU model to obtain a trained GRU model corresponding to each IMF component. Use the test sets of each IMF component to use the trained GRU model corresponding to each IMF component to obtain a predicted value for each IMF component. Superimpose the predicted values ​​of each IMF component to reconstruct a normalized internal resistance series predicted value. Denormalize the normalized internal resistance series predicted value using Formula 2 to obtain a battery state of health (SOH) predicted value. Formula 2: in, represents the battery health status predicted at the tth cycle, represents the normalized internal resistance sequence predicted at the tth cycle; Performing an end-of-life determination on the lithium-ion battery based on the predicted SOH value; if the predicted SOH value is less than or equal to a preset threshold, it indicates that the lithium-ion battery has reached the end of life, and marking the current cycle number as the end-of-life cycle number; If the SOH prediction value is greater than a preset threshold, it indicates that the current lithium-ion battery is in a healthy state.

5. A lithium-ion battery life prediction system based on variational mode decomposition (VMD), characterized by include: A data module is used to obtain a data set of lithium-ion battery charge and discharge cycles, obtain key health factors based on the data set, and normalize the key health factors; A decomposition module is used to decompose the normalized key health factors into several intrinsic mode function (IMF) components using the VMD method; The prediction module is used to construct a corresponding gated recurrent unit neural network GRU model for each IMF component, and use the corresponding GRU model to obtain a predicted value of each IMF component, and predict the end of the life of the lithium-ion battery based on the predicted value of each IMF component.

6. The system according to claim 5, characterized in that The data module is specifically used for: Extracting a discharge residual capacity sequence for each cycle based on the data set, and calculating a battery internal resistance sequence for each cycle based on the charge and discharge curves in the data set, verifying a strong correlation between the discharge residual capacity sequence and the battery internal resistance sequence according to a Pearson correlation coefficient, using the verified battery internal resistance sequence as a key health factor, normalizing the key health factor, and converting the normalized key health factor into a lithium-ion battery state of health (SOH) using Formula 1; Formula 1: in, Indicates the health status of the battery at the tth cycle, represents the true normalized internal resistance of the t-th cycle.

7. The system according to claim 5, characterized in that The decomposition module is specifically used for: According to the decomposition requirements, the normalized key health factors are decomposed into several intrinsic mode function (IMF) components using the VMD method; Among them, each IMF component is distributed from high to low according to frequency.

8. The system according to claim 5, characterized in that The prediction module is specifically used for: Divide the data set for each IMF component to generate respective training sets and test sets. Use the training sets of each IMF component to independently train a GRU model to obtain a trained GRU model corresponding to each IMF component. Use the test sets of each IMF component to use the trained GRU model corresponding to each IMF component to obtain a predicted value for each IMF component. Superimpose the predicted values ​​of each IMF component to reconstruct a normalized internal resistance series predicted value. Denormalize the normalized internal resistance series predicted value using Formula 2 to obtain a battery state of health (SOH) predicted value. Formula 2: in, represents the battery health status predicted at the tth cycle, represents the normalized internal resistance sequence predicted at the tth cycle; Performing an end-of-life determination on the lithium-ion battery based on the predicted SOH value; if the predicted SOH value is less than or equal to a preset threshold, it indicates that the lithium-ion battery has reached the end of life, and marking the current cycle number as the end-of-life cycle number; If the SOH prediction value is greater than a preset threshold, it indicates that the current lithium-ion battery is in a healthy state.

9. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the computer program is executed by the processor, the steps of the lithium-ion battery life prediction method based on variational mode decomposition (VMD) as described in any one of claims 1 to 4 are implemented.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores an information transmission implementation program, and when the program is executed by the processor, the steps of the lithium-ion battery life prediction method based on variational mode decomposition (VMD) as claimed in any one of claims 1 to 4 are implemented.

Images