Lithium ion battery health state prediction method and system based on health factor and KPCA-VMD-BILSTM

Abstract

The invention discloses a lithium ion battery health state prediction method and system based on a health factor and KPCA-VMD-BILSTM, and belongs to the technical field of lithium ion battery health state prediction. The prediction method comprises the following steps: collecting a plurality of health factors for predicting the SOH of the lithium ion battery, fusing the plurality of health factors for predicting the SOH of the lithium ion battery by using KPCA, and constructing a fused health index; performing multi-scale decomposition on the fusion health index by adopting VMD to obtain a plurality of modal components, and screening out the modal component having the strongest correlation with the SOH through correlation analysis; the SOH of the lithium ion battery is predicted on the screened modal component by using the trained BILSTM model, so that the time sequence dependency relationship in the battery degradation process can be captured, and the prediction precision is further improved; the method can be suitable for various types of lithium ion batteries, and has relatively high universality and practicability.

Description

Technical Field

[0001] The present invention belongs to the technical field of predicting the state of health of lithium-ion batteries, and particularly relates to a method and system for predicting the state of health of lithium-ion batteries based on health factors and KPCA-VMD-BILSTM. Background Art

[0002] Due to advantages such as high energy density, long cycle life, and low self-discharge rate, lithium-ion batteries are widely used in fields such as electric vehicles, energy storage systems, and aerospace. However, as the battery is used, its performance gradually degrades, and the state of health (SOH) decreases, directly affecting the safety and reliability of the system. Therefore, accurately predicting the SOH of lithium-ion batteries is of great significance for extending battery life and optimizing battery management.

[0003] Currently, traditional SOH prediction methods mainly rely on single health factors or simple statistical models, and it is difficult to comprehensively reflect the degradation characteristics of the battery. In addition, there are multi-scale characteristics such as global trends, local fluctuations, and noise interference during the battery degradation process, and traditional methods cannot effectively extract and utilize this information. Therefore, there is an urgent need for a method for predicting the SOH of lithium-ion batteries that can fuse multi-scale features and achieve high-precision prediction. Summary of the Invention

[0004] Aiming at the deficiencies of the existing technology, the purpose of the present invention is to provide a method and system for predicting the state of health of lithium-ion batteries based on health factors and KPCA-VMD-BILSTM, which solves the problems in the existing technology.

[0005] The purpose of the present invention can be achieved through the following technical solutions:

[0006] A method for predicting the state of health of lithium-ion batteries based on health factors and KPCA-VMD-BILSTM includes the following steps:

[0007] Collect multiple health factors for predicting the SOH of lithium-ion batteries, and use KPCA to fuse multiple health factors for predicting the SOH of lithium-ion batteries to construct a fused health index;

[0008] Use VMD to perform multi-scale decomposition on the fused health index to obtain multiple modal components, and screen out the modal component with the strongest correlation with SOH through correlation analysis;

[0009] Use the trained BILSTM model to predict the SOH of lithium-ion batteries for the screened modal components.

[0010] Furthermore, the health factors for predicting the SOH of lithium-ion batteries include: charge voltage rise time, charge voltage change, discharge voltage drop time, and discharge voltage change.

[0011] Further, the steps of constructing the fused health index include:

[0012] S11, taking the charge voltage rise time, charge voltage change, discharge voltage drop time, and discharge voltage change as the input layer data of KPCA, setting \(x = [x_1,...,x k \), and performing normalization processing on the data;

[0013] S12, using the Gaussian kernel function to map the normalized data into a high-dimensional space, and calculating the kernel matrix K;

[0014] S13, performing eigenvalue decomposition on the kernel matrix K, calculating the centralized kernel matrix C of the kernel matrix K, and extracting the principal components;

[0015] S14, linearly combining the extracted principal components with their corresponding variance contribution degrees as weights to form a fused health index.

[0016] Further, the calculation formula of the fused health index is:

[0017]

[0018] where FHI is the fused health index, \(\omega i is the variance contribution degree of the \(i\)-th Zhu Chengfeng, PC i is the \(i\)-th principal component, and \(n\) is the number of data.

[0019] Further, perform multi-scale decomposition on the fused health index using VMD, and the obtained modal components include: global change trend modal component, local regeneration modal component, and other noise modal components.

[0020] Further, when screening the modal component with the strongest correlation with SOH, by calculating the Person, Spearman, and Kendall correlation coefficients between the global change trend modal component, local regeneration modal component, other noise modal components and SOH, it is analyzed that the modal component with the strongest correlation with SOH is the global change trend modal component.

[0021] A lithium-ion battery health state prediction system based on health factors and KPCA-VMD-BILSTM includes:

[0022] Index construction module: Collect multiple health factors for predicting the SOH of lithium-ion batteries, and use KPCA to fuse multiple health factors for predicting the SOH of lithium-ion batteries to construct a fused health index;

[0023] Index decomposition module: Perform multi-scale decomposition on the fused health index using VMD to obtain multiple modal components, and screen out the modal component with the strongest correlation with SOH through correlation analysis;

[0024] And, a prediction module: using the trained BILSTM model to predict the state of health (SOH) of the lithium-ion battery for the selected modal components.

[0025] A computer storage medium stores a readable program that, when run by a processor, can execute the above load prediction method.

[0026] An electronic device includes: a processor, a memory, a communication interface, and a communication bus. The processor, the memory, and the communication interface complete communication with each other through the communication bus;

[0027] The memory is used to store at least one executable instruction, and the executable instruction causes the processor to execute the operations corresponding to the above load prediction method.

[0028] A computer program product includes computer instructions that direct a computing device to execute the operations corresponding to the above load prediction method.

[0029] Advantages of the present invention:

[0030] The present invention comprehensively reflects the degradation characteristics of the battery by fusing multiple health factors through kernel principal component analysis (KPCA), and then effectively extracts the global trend and local fluctuation characteristics during the battery degradation process by performing multi-scale decomposition on the fused health indicators using variational mode decomposition (VMD). Thus, prediction is carried out using a bidirectional long short-term memory network (BILSTM), which can capture the temporal dependence relationship during the battery degradation process and further improve the prediction accuracy; this method is applicable to various types of lithium-ion batteries and has strong versatility and practicality. Description of the Drawings

[0031] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0032] Figure 1 is the flowchart of the state of health prediction of the lithium-ion battery of the present invention;

[0033] Figure 2 is the change curve of the charging voltage rise time at different cycle periods in the structural schematic diagram of the present invention;

[0034] Figure 3 is the schematic diagram of VMD multi-scale decomposition;

[0035] Figure 4 is the schematic diagram of the BILSTM network structure;

[0036] Figure 5 is the SOH prediction diagram of the lithium-ion battery of the present invention. Specific embodiments

[0037] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0038] Embodiment 1

[0039] As Figure 1 shown, a method for predicting the health state of a lithium-ion battery based on a health factor and KPCA-VMD-BILSTM includes the following steps:

[0040] S1, collect multiple health factors for predicting the SOH of the lithium-ion battery, and use KPCA to fuse multiple health factors for predicting the SOH of the lithium-ion battery to construct a fused health index (FHI); wherein, the health factors for predicting the SOH of the lithium-ion battery include: the charging voltage rise time (HI1) of the lithium-ion battery, the charging voltage change (HI2), the discharge voltage drop time (HI3), and the discharge voltage change (HI4);

[0041] In this embodiment, the experimental data of the lithium-ion battery is from the public dataset provided by NASA's PCoE. Select the charge and discharge data of three 18650-type lithium-ion batteries (B5, B6, B7) collected under the same experimental environment, and the rated capacity of the battery is 2 A·h; obtain the charge and discharge data of the lithium-ion battery from the above public dataset, and extract the charging voltage rise time (HI1), charging voltage change (HI2), discharge voltage drop time (HI3), and discharge voltage change (HI4) of the lithium-ion battery.

[0042] In this embodiment, by calculating the Person, Spearman, and Kendall correlation coefficients between the health factor and the SOH of the lithium-ion battery, health factors with relatively high correlations are selected, as shown in Table 1;

[0043] Table 1 Correlation analysis table of health factors and SOH (taking battery B5 as an example)

[0044]

[0045] The charging voltage rise time (HI1), charging voltage change (HI2), discharging voltage drop time (HI3), and discharging voltage change (HI4) have relatively high correlation coefficients with the state of health SOH obtained from the calculation results. Therefore, they can be used as health factors to reflect the SOH of lithium-ion batteries.

[0046] Taking the charging voltage rise time (HI1) as an example, Figure 2 The change curve of the charging voltage rise time for different cycle periods is given. It can be seen that as the number of cycles increases, the overall voltage curve moves forward, and the time from the same voltage starting point to the cut-off voltage becomes shorter and shorter, showing a significant change.

[0047] The steps to construct the fused health index (FHI) include:

[0048] S11, taking the charging voltage rise time (HI1), charging voltage change (HI2), discharging voltage drop time (HI3), and discharging voltage change (HI4) as the input layer data of KPCA and setting it as x = [x1,...,x k , and performing standardization processing on the data to eliminate missing values and outliers in the data;

[0049] The formula for standardization processing is:

[0050]

[0051] where x' is the standardized data, μ is the mean value, and σ is the standard deviation.

[0052] S12, using the Gaussian kernel function to map the standardized data to a high-dimensional space and calculating the kernel matrix K; the calculation formula is:

[0053]

[0054] where each element K in the kernel matrix K ij represents the inner product of the i-th sample and the j-th sample in the high-dimensional space;

[0055] S13, performing eigenvalue decomposition on the kernel matrix K, calculating the centralized kernel matrix C of the kernel matrix K, and extracting the principal component PC i ; the calculation formula is:

[0056] C = K - K·K 均匀 ·K

[0057] In the formula, K 均匀 is the uniform matrix.

[0058] S14, linearly combining the extracted principal components with their corresponding variance contribution degrees as weights to form a fused health index, and the specific expression is:

[0059]

[0060] Among them, FHI is the fused health index, ω i is the variance contribution degree of the i-th Zhu Chengfeng, PC i is the i-th principal component, and n is the number of data.

[0061] S2. Use VMD to perform multi-scale decomposition on the fused health index (FHI) to obtain multiple modal components, and screen out the modal component with the strongest correlation with SOH through correlation analysis;

[0062] The decomposed modal components include: the global change trend modal component (IMF1), the local regeneration modal component (IMF2), and other noise modal components (IMF3);

[0063] As Figure 3 shown, the process of using VMD to perform multi-scale decomposition on the fused health index includes:

[0064] Denote the health index (FHI) after KPCA fusion as the signal f(t); define that the signal f(t) is decomposed into N modal components IMF by VMD, and each modal component is represented by IMF N ; its variational model is as follows:

[0065]

[0066] Among them, ω n is the central frequency of the N-th modal component, represents the reciprocal of time, δ(t) is the Dirac function, * represents the convolution operation, u n (t) is the N-th modal component function, j is the imaginary unit, f(t) is the input signal of the fused health index, and N is the total number of decomposed modal components;

[0067] After the variational mode decomposition VMD iteration, the signal f(t) can be decomposed into three modal components: global trend, local regeneration, and other noise. Therefore, set N = 3; set the lowest central frequency of 0 - 0.1 Hz, which reflects the long-term change trend characteristics of the fused health index and represents the global trend of the battery health state, denoted as IMF1; set the central frequency range of 0.1 - 1 Hz, which reflects the local regeneration characteristics and represents the periodic regeneration fluctuation of the battery health state, denoted as IMF2; set the high-frequency component greater than 1 Hz, which is determined as the noise-dominated mode and mainly contains the noise components in the data, denoted as IMF3.

[0068] The steps to screen out the modal component with the strongest correlation with SOH include:

[0069] 1) Calculate the Pearson, Spearman, and Kendall correlation coefficients between the global trend mode component (IMF1), local regenerative mode component (IMF2), other noise mode components (IMF3), and SOH.

[0070] 2) In this embodiment, through correlation analysis, it is obtained that there is the strongest correlation between IMF1 and SOH. Therefore, IMF1 is selected as the main input for subsequent prediction.

[0071] S3. Use the trained BILSTM model to accurately predict the SOH of the lithium-ion battery for the mode components selected in S2.

[0072] As Figure 4 shown, the BILSTM includes: The BILSTM consists of two LSTM sub-networks: LSTM+ is the forward LSTM, and LSTM- is the backward LSTM; ht+ is the hidden state from the forward LSTM, representing the information from the past to the current time, and ht- is the hidden state from the backward LSTM, representing the information from the future to the current time.

[0073] The input data x t is sent into the forward LSTM and the backward LSTM simultaneously. The forward LSTM processes the sequence forward, learns the time-dependent features of the current time step and the information before it, captures the past context relationship, and calculates the hidden state sequence ht+; the backward LSTM processes the sequence backward, learns the time-dependent features of the current time step and the information after it, captures the future context relationship, and calculates the hidden state sequence ht-.

[0074] For each time step t, ht+ and ht- are fused to obtain the final output ht of this time step, and this output contains the bidirectional information of the context.

[0075] The prediction process includes:

[0076] S31. Divide the mode component with the strongest correlation (IMF1) and the real SOH data of the lithium-ion battery into a training set and a test set according to a certain ratio, and use the training set to train the BILSTM model to optimize the model parameters.

[0077] S32. Use the trained BILSTM model to predict the test set to obtain the prediction result of the battery health state SOH.

[0078] Taking B5 as an example, Figure 5 the SOH prediction diagram of the lithium-ion battery is given; it can be seen from the figure that...

[0079] During multiple charge-discharge cycles of a lithium-ion battery, the curve of the KPCA-VMD-BILSTM prediction model highly coincides with the curve of the true SOH value. Especially in the middle and late stages, from the 60th to the 160th cycle, the prediction results still closely follow the true SOH, indicating that the model has a high prediction accuracy for SOH. It can effectively capture the overall trend and local fluctuation characteristics of battery degradation, achieving high-precision dynamic prediction of the SOH of lithium-ion batteries in long-time series prediction, and having good practical application potential.

[0080] Based on a similar inventive concept, an embodiment of the present invention further provides a computer storage medium storing a readable program that, when run by a processor, can execute the above-mentioned method for predicting the health state of a lithium-ion battery based on health factors and KPCA-VMD-BILSTM.

[0081] Based on a similar inventive concept, an embodiment of the present invention provides an electronic device, including: a processor, a memory, a communication interface, and a communication bus, where the processor, the memory, and the communication interface complete communication with each other through the communication bus;

[0082] The memory is used to store at least one executable instruction, and the executable instruction causes the processor to execute the operations corresponding to the above-mentioned method for predicting the health state of a lithium-ion battery based on health factors and KPCA-VMD-BILSTM.

[0083] Based on a similar inventive concept, an embodiment of the present invention further provides a computer program product including computer instructions that direct a computing device to execute the operations corresponding to the above-mentioned method for predicting the health state of a lithium-ion battery based on health factors and KPCA-VMD-BILSTM.

[0084] Embodiment 2

[0085] In this embodiment, a prediction system for the health state of a lithium-ion battery based on health factors and KPCA-VMD-BILSTM is proposed, including:

[0086] An index construction module: collecting multiple health factors for predicting the SOH of a lithium-ion battery, and using KPCA to fuse multiple health factors for predicting the SOH of a lithium-ion battery to construct a fused health index;

[0087] An index decomposition module: performing multi-scale decomposition on the fused health index using VMD to obtain multiple modal components, and screening out the modal component with the strongest correlation with SOH through correlation analysis;

[0088] And a prediction module: using the trained BILSTM model to predict the SOH of the lithium-ion battery for the screened modal component.

[0089] The method of the present invention can be implemented in hardware, firmware, or be implemented as software or computer code that can be stored in a recording medium (such as a CDROM, RAM, floppy disk, hard disk, or magneto-optical disk), or be implemented as computer code originally stored in a remote recording medium or a non-transitory machine-readable medium and to be downloaded through a network and stored in a local recording medium, so that the method described herein can be stored on such a software process on a recording medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware (such as an ASIC or FPGA). It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component (such as RAM, ROM, flash memory, etc.) that can store or receive software or computer code, and when the software or computer code is accessed and executed by the computer, the processor, or the hardware, the method described herein is implemented. In addition, when a general-purpose computer accesses the code for implementing the method shown herein, the execution of the code converts the general-purpose computer into a dedicated computer for executing the method shown herein.

[0090] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art of this industry should understand that the present invention is not limited by the above embodiments. What is described in the above embodiments and the specification only illustrates the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed.

Claims

1. A method for predicting the state of health of a lithium-ion battery based on health factors and KPCA-VMD-BILSTM, characterized in that, It includes the following steps: Collect multiple health factors for predicting the state of health (SOH) of a lithium-ion battery, and use kernel principal component analysis (KPCA) to fuse the multiple health factors for predicting the SOH of the lithium-ion battery to construct a fused health index; Use variational mode decomposition (VMD) to perform multi-scale decomposition on the fused health index to obtain multiple modal components, and screen out the modal component with the strongest correlation with the SOH through correlation analysis; Use the trained bidirectional long short-term memory (BILSTM) model to predict the SOH of the lithium-ion battery for the screened modal component.

2. The method for predicting the health state of a lithium-ion battery based on health factors and KPCA-VMD-BILSTM according to claim 1, wherein The health factors for predicting the SOH of the lithium-ion battery include: charging voltage rise time, charging voltage change, discharging voltage drop time, and discharging voltage change.

3. The method for predicting the health state of a lithium-ion battery based on health factors and KPCA-VMD-BILSTM according to claim 2, wherein, The steps for constructing the fused health index include: S11, Set the charge voltage rise time, charge voltage change, discharge voltage drop time, and discharge voltage change as the input layer data of KPCA to x = [x1,..., x k , and perform standardization processing on the data; S12, use the Gaussian kernel function to map the standardized data to a high-dimensional space, and calculate the kernel matrix K; S13, perform eigen-decomposition on the kernel matrix K, calculate the centralized kernel matrix C of the kernel matrix K, and extract the principal components; S14, perform a linear combination of the extracted principal components with their corresponding variance contribution degrees as weights to form a fused health index.

4. The method for predicting the health state of a lithium-ion battery based on health factors and KPCA-VMD-BILSTM according to claim 3, characterized in that, The calculation formula of the fused health index is: Among them, FHI is the fused health index, ω i is the variance contribution degree of the i-th Zhu Chengfeng, PC i is the i-th principal component, and n is the number of data points.

5. The method for predicting the health state of a lithium-ion battery based on health factors and KPCA-VMD-BILSTM according to claim 1, wherein Using VMD to perform multi-scale decomposition on the fused health index, the obtained modal components include: global change trend modal component, local regeneration modal component, and other noise modal components.

6. The method for predicting the health state of a lithium-ion battery based on health factors and KPCA-VMD-BILSTM according to claim 5, characterized in that When screening the modal component with the strongest correlation with the SOH, by calculating the Pearson, Spearman, and Kendall correlation coefficients between the global change trend modal component, local regeneration modal component, other noise modal components and the SOH, it is analyzed that the modal component with the strongest correlation with the SOH is the global change trend modal component.

7. A lithium-ion battery health state prediction system based on health factors and KPCA-VMD-BILSTM, characterized in that, It includes: Index construction module: Collect multiple health factors for predicting the SOH of the lithium-ion battery, and use KPCA to fuse the multiple health factors for predicting the SOH of the lithium-ion battery to construct a fused health index; Index decomposition module: Use VMD to perform multi-scale decomposition on the fused health index to obtain multiple modal components, and screen out the modal component with the strongest correlation with the SOH through correlation analysis; And a prediction module: Use the trained BILSTM model to predict the SOH of the lithium-ion battery for the screened modal component.

8. A computer storage medium stores a readable program, characterized in that, When the program is run by a processor, it can execute the load prediction method described in any one of claims 1-6.

9. An electronic device, characterized in that, It includes: A processor, a memory, a communication interface, and a communication bus. The processor, the memory, and the communication interface complete communication with each other through the communication bus; The memory is used to store at least one executable instruction, and the executable instruction causes the processor to execute the operations corresponding to the load prediction method described in any one of claims 1-6.

10. A computer program product comprising computer instructions, characterized in that, The computer instruction instructs the computing device to execute the operations corresponding to the load prediction method described in any one of claims 1-6.

Images