Battery loss prediction method, system and equipment based on deep learning, and medium

Abstract

The invention discloses a battery loss prediction method, system and device based on deep learning and a medium, and relates to the technical field of agricultural machinery, and the method comprises the steps: determining the operation data of a to-be-predicted battery under different working conditions; decomposing the operation data to obtain modal functions of different frequency components, determining a time domain feature and a frequency domain feature corresponding to the modal function under each frequency component, and integrating all the time domain features and all the frequency domain features to generate a feature vector; and processing the feature vector according to an LSTM model to obtain a battery loss result corresponding to the to-be-predicted battery. According to the method, the precision and adaptability of new energy battery loss prediction can be effectively improved, the calculation complexity of the model is reduced, and the dependence on a large amount of historical data is reduced.

Description

Background Art

[0002] New energy batteries are an important part of the current energy field and are widely used in electric vehicles, energy storage systems, renewable energy generation, etc. With the increase in global demand for clean energy, the application scope and quantity of new energy batteries have expanded dramatically. However, batteries will gradually lose power during use, affecting their performance and life. Accurately predicting battery loss is of great significance for optimizing battery use, extending battery life, and improving economic benefits.

[0003] The loss of new energy batteries is mainly reflected in two aspects: capacity attenuation and increased internal resistance. Capacity attenuation refers to the gradual decrease in the amount of electricity that can be stored in the battery during the cycle; the increase in internal resistance leads to a decrease in the discharge efficiency of the battery and an increase in heat generation. Factors affecting battery loss include the number of charge and discharge cycles, charge and discharge rates, operating temperature and other factors. The discharge mass ratio of the battery (referred to as QD value) is one of the important indicators for measuring battery performance. The QD value represents the amount of electricity released by a unit mass of battery material during the discharge process, usually in units of mAh / g (milliampere-hours per gram). The sampling method is to take a certain amount of battery active material (such as positive electrode material or negative electrode material), evenly coat the material on a conductive substrate (such as aluminum foil or copper foil) to prepare an electrode sheet. On this basis, the prepared electrode sheet is assembled into a half-cell or full cell together with an electrolyte, a diaphragm, and a counter electrode (such as a lithium metal sheet or an activated carbon electrode). Use an electrochemical workstation or a battery testing system to perform charge and discharge tests on the assembled battery. Perform multiple cycle tests according to the preset charge and discharge current density. Record the amount of electricity (mAh) released by the battery during discharge, and calculate the QD value based on the total amount of electricity released during discharge and the mass of the electrode material. The calculation method is:

[0004]

[0005] At present, the methods for predicting battery loss mainly include statistical models based on experimental data, models based on physical mechanisms, and data-driven machine learning models.

[0006] Statistical model based on experimental data: Through the accumulation and analysis of a large amount of experimental data, an empirical model of battery loss is established. The advantages of this type of method are simplicity and intuitiveness, but the disadvantages are that a large amount of experimental data is required and the generalization ability of the model is limited.

[0007] Models based on physical mechanisms: By analyzing the chemical reactions and physical changes inside the battery, a loss prediction model is established. For example, models based on the growth of solid electrolyte interface films, models based on changes in electrode material structure, etc. The advantage of this type of method is that the model is highly interpretable, but it requires a deep understanding of the internal mechanism of the battery and the model is relatively complex.

[0008] Data-driven machine learning model: Use machine learning algorithms to extract features and patterns from historical data and establish a battery loss prediction model. Common methods include support vector machines, random forests, neural networks, etc. The advantage of this type of method is that it can handle complex nonlinear relationships and has high prediction accuracy. The disadvantage is that it requires a large amount of historical data for training. Summary of the invention

[0009] The technical problem to be solved by the present invention is to address the shortcomings of the prior art, specifically the problems of slow development cycle and long debugging time, and specifically provide a battery loss prediction method, system, device and medium based on deep learning, as follows:

[0010] 1) In the first aspect, the present invention provides a battery loss prediction method based on deep learning, and the specific technical solution is as follows:

[0011] S1, determining the operating data of the battery to be predicted under different working conditions;

[0012] S2, decomposing the operating data to obtain modal functions of different frequency components, determining the time domain features and frequency domain features corresponding to the modal functions under each frequency component, integrating all the time domain features and all the frequency domain features to generate a feature vector;

[0013] S3, processing the feature vector according to the LSTM model to obtain the battery loss result corresponding to the battery to be predicted.

[0014] The beneficial effects of a battery loss prediction method based on deep learning provided by the present invention are as follows:

[0015] It can effectively improve the accuracy and adaptability of new energy battery loss prediction, reduce the computational complexity of the model, and reduce dependence on large amounts of historical data.

[0016] Based on the above solution, the present invention can also be improved as follows.

[0017] Furthermore, the process of decomposing the operating data to obtain modal functions of different frequency components is specifically as follows:

[0018] Decomposing the operating data by using an optimized variational modal decomposition method to obtain modal functions of different frequency components;

[0019] The optimized variational mode decomposition method is a variational mode decomposition method after parameter optimization, and the parameter optimization process is specifically as follows:

[0020] Determine multiple training mode functions corresponding to the training operation data, and determine the center frequency corresponding to each training mode function, and by combining constraints, determine that the sum of the multiple training mode functions is equal to the original signal, and use the parameters of the variational mode decomposition method corresponding to the sum of the multiple training mode functions being equal to the original signal as the optimal parameters, and the variational mode decomposition method determined based on the optimal parameters is the optimized variational mode decomposition method.

[0021] Furthermore, the LSTM model consists of an input layer, an LSTM layer, a fully connected layer and an output layer;

[0022] The LSTM layer is composed of multiple neural units, each of which includes a forget gate, an input gate, a memory cell, and an output gate.

[0023] Furthermore, it also includes:

[0024] Preprocessing the operation data to obtain preprocessed operation data, using the preprocessed operation data as the operation data in S2 and executing S2;

[0025] The preprocessing includes: cleaning and normalizing the operation data.

[0026] 2) In the second aspect, the present invention also provides a battery loss prediction system based on deep learning, and the specific technical solution is as follows:

[0027] The determination module is used to: determine the operating data of the battery to be predicted under different working conditions;

[0028] The decomposition module is used to: decompose the operating data to obtain modal functions of different frequency components, determine the time domain features and frequency domain features corresponding to the modal function under each frequency component, integrate all the time domain features and all the frequency domain features, and generate a feature vector;

[0029] The prediction module is used to: process the feature vector according to the LSTM model to obtain the battery loss result corresponding to the battery to be predicted.

[0030] Based on the above solution, the present invention can also be improved as follows.

[0031] Furthermore, the process of decomposing the operating data to obtain modal functions of different frequency components is specifically as follows:

[0032] Decomposing the operating data by using an optimized variational modal decomposition method to obtain modal functions of different frequency components;

[0033] The optimized variational mode decomposition method is a variational mode decomposition method after parameter optimization, and the parameter optimization process is specifically as follows:

[0034] Determine multiple training mode functions corresponding to the training operation data, and determine the center frequency corresponding to each training mode function, and by combining constraints, determine that the sum of the multiple training mode functions is equal to the original signal, and use the parameters of the variational mode decomposition method corresponding to the sum of the multiple training mode functions being equal to the original signal as the optimal parameters, and the variational mode decomposition method determined based on the optimal parameters is the optimized variational mode decomposition method.

[0035] Furthermore, the LSTM model consists of an input layer, an LSTM layer, a fully connected layer and an output layer;

[0036] The LSTM layer is composed of multiple neural units, each of which includes a forget gate, an input gate, a memory cell, and an output gate.

[0037] Furthermore, it also includes:

[0038] The preprocessing module is used to: preprocess the operation data to obtain preprocessed operation data, use the preprocessed operation data as the operation data in the decomposition module and run the decomposition module;

[0039] The preprocessing includes: cleaning and normalizing the operation data.

[0040] 3) In a third aspect, the present invention further provides an electronic device, comprising a processor, wherein the processor is coupled to a memory, wherein the memory stores at least one computer program, and wherein the at least one computer program is loaded and executed by the processor so that the electronic device implements any of the above methods.

[0041] 4) In a fourth aspect, the present invention further provides a computer-readable storage medium, wherein at least one computer program is stored in the computer-readable storage medium, and the at least one computer program is loaded and executed by a processor so that a computer implements any of the above methods.

[0042] It should be noted that the beneficial effects achieved by the technical solutions of the second to fourth aspects of the present invention and the corresponding possible implementation methods can be found in the above-mentioned technical effects of the first aspect and its corresponding possible implementation methods, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:

[0044] Figure 1 A schematic diagram of a flow chart of a battery loss prediction method based on deep learning according to an embodiment of the present invention;

[0045] Figure 2 A data processing flow chart of a battery loss prediction method based on deep learning according to an embodiment of the present invention;

[0046] Figure 3 A system architecture diagram of a battery loss prediction method based on deep learning according to an embodiment of the present invention;

[0047] Figure 4 A schematic diagram of the LSTM model structure of a battery loss prediction method based on deep learning according to an embodiment of the present invention;

[0048] Figure 5 A schematic diagram of a parameter fusion structure of a battery loss prediction method based on deep learning according to an embodiment of the present invention;

[0049] Figure 6 A schematic diagram of a model training process of a battery loss prediction method based on deep learning according to an embodiment of the present invention;

[0050] Figure 7 A schematic diagram of LSTM algorithm prediction results of a battery loss prediction method based on deep learning according to an embodiment of the present invention;

[0051] Figure 8 A schematic diagram of prediction results of a VMD+LSTM algorithm of a battery loss prediction method based on deep learning according to an embodiment of the present invention;

[0052] Fig. 9 The present invention is a structural framework diagram of an electronic device. DETAILED DESCRIPTION

[0053] To make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0054] like Figures 1 to 8 As shown, a battery loss prediction method based on deep learning in an embodiment of the present invention includes the following steps:

[0055] S1, determining the operating data of the battery to be predicted under different working conditions;

[0056] S2, decomposing the operating data to obtain modal functions of different frequency components, determining the time domain features and frequency domain features corresponding to the modal functions under each frequency component, integrating all the time domain features and all the frequency domain features to generate a feature vector;

[0057] S3, processing the feature vector according to the LSTM model to obtain the battery loss result corresponding to the battery to be predicted.

[0058] The beneficial effects of a battery loss prediction method based on deep learning provided by the present invention are as follows:

[0059] It can effectively improve the accuracy and adaptability of new energy battery loss prediction, reduce the computational complexity of the model, and reduce dependence on large amounts of historical data.

[0060] S1, in the process of determining the operating data of the battery to be predicted under different working conditions:

[0061] Different working conditions refer to at least one of: working in a low temperature environment, working in a high temperature environment, and normal working.

[0062] Operation data includes but is not limited to: time series data such as voltage, current, temperature, etc. Sensors are used to monitor various parameters of the battery in real time.

[0063] The above process is specifically as follows: collect the operating data of the battery under different working conditions. That is, use sensors to monitor various parameters of the battery in real time and record them through the data acquisition system. The data is stored in the local database for subsequent processing.

[0064] S2, decomposing the operating data to obtain modal functions of different frequency components, determining the time domain features and frequency domain features corresponding to the modal functions under each frequency component, integrating all the time domain features and all the frequency domain features, and generating a feature vector:

[0065] The variational mode decomposition (VMD) method is used to decompose the time series data and extract the mode functions of different frequency components.

[0066] It should be noted that the VMD algorithm decomposes the input time series signal into several intrinsic mode functions (IMF), each of which represents a signal with a specific frequency component. The decomposition method of the battery charge curve in the deep learning-based battery loss prediction is as follows: Step 1. Process the original power load data to identify local maximum points and local minimum points. Step 2. Use the cubic spline interpolation method to construct the upper envelope and the lower envelope, where the upper envelope is composed of all local maximum points and the lower envelope is composed of all local minimum points, and the average value of the two envelopes is calculated. Step 3. The difference value is obtained by calculating the difference between the load data and the average value of the envelope, and the difference value is evaluated to confirm whether it meets the constraints. If not, repeat steps 1 to 3 until all constraints are met.

[0067] In order to ensure the calculation accuracy of the VMD algorithm, the VMD algorithm can be optimized to obtain the optimized VMD algorithm. The optimization process is parameter optimization, and the specific implementation method is:

[0068] Parameter optimization: Determine the optimal parameters of the VMD algorithm, such as the number of decomposition layers, penalty parameters, etc., through methods such as cross-validation. The goal of VMD is to decompose the signal f(t) into K modes u k (t), each mode corresponds to a center frequency w k , the goal of finding the optimal parameters for the VMD algorithm is achieved by solving the following variational problem:

[0069]

[0070] where u k (t) is the kth mode, w k is the center frequency of the kth mode, δ(t) is the Dirac function, j is the imaginary unit, * represents the convolution operation, represents the time derivative, K represents the number of modes, and t represents the time. It ensures that the sum of all modes is equal to the original signal through the following constraints:

[0071]

[0072] To facilitate the solution of variational problems, the VMD algorithm introduces Lagrangian multipliers and penalty factors to obtain the Lagrangian function:

[0073]

[0074] The Lagrangian function is used to update the function alternately, thus approaching the optimal solution. The iterative process is: k 、w k , λ initialization, u k represents the kth modal function component, w k represents the center frequency of the kth basis function, λ represents the Lagrange multiplication operator, and α represents the Dirac function, which is continuously updated. and λ n+1 , until certain iteration constraints are met:

[0075]

[0076] in represents the basis functions for obtaining new modal components, represents the center frequency of each new basis function, λ n+1 (w) represents the new Lagrange multiplier solution that satisfies the constraints and minimizes the objective function, w represents the center frequency, f(w) represents the objective function at the center frequency, u i (w) represents the function modal component at the center frequency, λ(w) represents the Lagrange multiplier solution, and w k represents the kth center frequency, and τ represents the penalty parameter.

[0077] Feature extraction is performed on the decomposed modal function to generate a feature vector suitable for LSTM model input.

[0078] Feature selection: Extract time domain features (such as mean, variance, kurtosis, etc.) and frequency domain features (such as spectral energy, main frequency, etc.) from the decomposed modal function.

[0079] Feature combination: The extracted features are combined into a feature vector as the input of the LSTM model.

[0080] S3, according to the LSTM model, the feature vector is processed to obtain the battery loss result corresponding to the battery to be predicted:

[0081] Use long short-term memory networks for model training and prediction to achieve high-precision prediction of battery loss. LSTM model construction: Design a deep learning model that includes an input layer, an LSTM layer, a fully connected layer, and an output layer. Model training: Use the extracted feature vectors and the corresponding battery loss labels for supervised learning, and optimize the model parameters through the back-propagation algorithm. Prediction and reasoning: Input new battery data into the trained LSTM model for loss prediction. The LSTM layer consists of multiple neural units. The internal structure of each neuron in the LSTM layer consists of four parts: a forget gate, an input gate, a memory cell, and an output gate:

[0082]

[0083] where f t is the forgetting coefficient, i t is the input coefficient, is the input data, C t is the updated cell state, o t is the output coefficient, h t is the output data, σ is the sigmoid activation function. f , W i , W c and W o are the forget gate weight, input gate weight, input data weight and output gate weight respectively, b f , b i , b c and b o They are forget gate bias, input gate bias, input data bias and output gate bias respectively.

[0084] Also includes:

[0085] Evaluate the prediction results of the model, provide feedback on the model performance and make corresponding adjustments. Evaluation indicators: use mean square error (MSE), mean absolute error (MAE), coefficient of determination (R 2) and other indicators to evaluate the model prediction performance. Model optimization: adjust the model structure and parameters according to the evaluation results to further improve the prediction accuracy and robustness. Denormalize the predicted values ​​obtained by each LSTM prediction model, and then perform aggregation operations to obtain the final actual predicted values. Introduce root mean square error (RMSE), mean absolute error (MAE), and determination coefficient (R-Square) indicators to test the prediction effect of the model.

[0086] The root mean square error (RMSE) can reflect the discreteness of the predicted value and the size of the model error.

[0087]

[0088] n represents the number of samples, y i Indicates the actual value, Represents the predicted value.

[0089] The mean absolute error (MAE) is the average of the absolute errors and can better reflect the actual situation of the prediction error.

[0090]

[0091] The coefficient of determination (R-Square) can reflect the degree of explanation of the independent variable on the dependent variable, and its value range is 0-1. The higher the degree of explanation, the closer the coefficient value is to 1.

[0092]

[0093] in, Represents the mean of the actual values.

[0094] Table 1 Evaluation index table

[0095] Model Proposed method EMD-LSTM Bi-LSTM LSTM <![CDATA[R 2 ]]> 0.917 0.697 0.833 0.611 RMSE 124.3 236.98 175.79 241.6 MAE 83.21 122.93 100.43 125.7

[0096] The technical solution of the present invention can effectively improve the accuracy and adaptability of new energy battery loss prediction, reduce the calculation complexity of the model, and reduce the dependence on a large amount of historical data, with the following specific advantages:

[0097] Improve prediction accuracy: VMD decomposition improves the expressiveness of data features, and the LSTM model captures complex temporal dependencies.

[0098] Enhanced model adaptability: Comprehensive signal processing and deep learning methods improve the model's adaptability under different operating conditions and battery types.

[0099] Reduce computational complexity: Through feature extraction and model optimization, the amount of computation is reduced and real-time performance is improved.

[0100] Reduce data dependence: The combination of VMD and LSTM improves the model's predictive ability under limited data and reduces the need for large-scale high-quality historical data.

[0101] In summary, the new energy battery loss prediction method based on VMD and LSTM proposed in the present invention provides effective technical support for the battery management system, which helps to extend the battery life and improve the economic benefits.

[0102] Furthermore, the process of decomposing the operating data to obtain modal functions of different frequency components is specifically as follows:

[0103] Decomposing the operating data by using an optimized variational modal decomposition method to obtain modal functions of different frequency components;

[0104] The optimized variational mode decomposition method is a variational mode decomposition method after parameter optimization, and the parameter optimization process is specifically as follows:

[0105] Determine multiple training mode functions corresponding to the training operation data, and determine the center frequency corresponding to each training mode function, and by combining constraints, determine that the sum of the multiple training mode functions is equal to the original signal, and use the parameters of the variational mode decomposition method corresponding to the sum of the multiple training mode functions being equal to the original signal as the optimal parameters, and the variational mode decomposition method determined based on the optimal parameters is the optimized variational mode decomposition method.

[0106] Furthermore, the LSTM model consists of an input layer, an LSTM layer, a fully connected layer and an output layer;

[0107] The LSTM layer is composed of multiple neural units, each of which includes a forget gate, an input gate, a memory cell, and an output gate.

[0108] Furthermore, it also includes:

[0109] Preprocessing the operation data to obtain preprocessed operation data, using the preprocessed operation data as the operation data in S2 and executing S2;

[0110] The preprocessing includes: cleaning and normalizing the operation data.

[0111] The present invention also provides a battery loss prediction system based on deep learning, and the specific technical solution is as follows:

[0112] The determination module is used to: determine the operating data of the battery to be predicted under different working conditions;

[0113] The decomposition module is used to: decompose the operating data to obtain modal functions of different frequency components, determine the time domain features and frequency domain features corresponding to the modal function under each frequency component, integrate all the time domain features and all the frequency domain features, and generate a feature vector;

[0114] The prediction module is used to: process the feature vector according to the LSTM model to obtain the battery loss result corresponding to the battery to be predicted.

[0115] Based on the above solution, the present invention can also be improved as follows.

[0116] Furthermore, the process of decomposing the operating data to obtain modal functions of different frequency components is specifically as follows:

[0117] Decomposing the operating data by using an optimized variational modal decomposition method to obtain modal functions of different frequency components;

[0118] The optimized variational mode decomposition method is a variational mode decomposition method after parameter optimization, and the parameter optimization process is specifically as follows:

[0119] Determine multiple training mode functions corresponding to the training operation data, and determine the center frequency corresponding to each training mode function, and by combining constraints, determine that the sum of the multiple training mode functions is equal to the original signal, and use the parameters of the variational mode decomposition method corresponding to the sum of the multiple training mode functions being equal to the original signal as the optimal parameters, and the variational mode decomposition method determined based on the optimal parameters is the optimized variational mode decomposition method.

[0120] Furthermore, the LSTM model consists of an input layer, an LSTM layer, a fully connected layer and an output layer;

[0121] The LSTM layer is composed of multiple neural units, each of which includes a forget gate, an input gate, a memory cell, and an output gate.

[0122] Furthermore, it also includes:

[0123] The preprocessing module is used to: preprocess the operation data to obtain preprocessed operation data, use the preprocessed operation data as the operation data in the decomposition module and run the decomposition module;

[0124] The preprocessing includes: cleaning and normalizing the operation data.

[0125] In the above embodiments, although the steps are numbered S1, S2, etc., these are only specific embodiments given by the present invention. Those skilled in the art may adjust the execution order of S1, S2, etc. according to actual conditions, which is also within the protection scope of the present invention. It can be understood that in some embodiments, some or all of the above embodiments may be included.

[0126] It should be noted that the beneficial effects of the battery loss prediction system based on deep learning provided by the above embodiment are the same as the beneficial effects of the battery loss prediction method based on deep learning, which will not be repeated here. In addition, when the system provided by the above embodiment realizes its functions, it only uses the division of the above functional modules as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the system is divided into different functional modules according to actual conditions to complete all or part of the functions described above. In addition, the system and method embodiments provided by the above embodiment belong to the same concept. The specific implementation process is detailed in the method embodiment, which will not be repeated here.

[0127] like Fig. 9 As shown, an electronic device 300 according to an embodiment of the present invention includes a processor 320, the processor 320 is coupled to a memory 310, and the memory 310 stores at least one computer program 330, and the at least one computer program 330 is loaded and executed by the processor 320, so that the electronic device 300 implements any of the above methods, specifically:

[0128] The electronic device 300 may have relatively large differences due to different configurations or performances, and may include one or more processors 320 (Central Processing Units, CPU) and one or more memories 310, wherein the one or more memories 310 store at least one computer program 330, and the at least one computer program 330 is loaded and executed by the one or more processors 320, so that the electronic device 300 implements a battery loss prediction method based on deep learning provided in the above embodiment. Of course, the electronic device 300 may also have components such as a wired or wireless network interface, a keyboard, and an input and output interface for input and output. The electronic device 300 may also include other components for realizing device functions, which will not be repeated here.

[0129] A computer-readable storage medium according to an embodiment of the present invention stores at least one computer program, and the at least one computer program is loaded and executed by a processor so that a computer implements any one of the above methods.

[0130] Optionally, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a compact disc (CD-ROM), a magnetic tape, a floppy disk, an optical data storage device, and the like.

[0131] In an exemplary embodiment, a computer program product or a computer program is also provided, the computer program product or the computer program comprising computer instructions, the computer instructions being stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the electronic device performs any of the above methods.

[0132] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects and represent the definition of a specific order or sequence. The order of use of similar objects can be interchanged where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than the order shown or described.

[0133] Those skilled in the art know that the present invention can be implemented as a system, method or computer program product. Therefore, the present disclosure can be specifically implemented in the following forms, namely: it can be complete hardware, it can be complete software (including firmware, resident software, microcode, etc.), or it can be a combination of hardware and software, which is generally referred to as "circuit", "module" or "system" herein. In addition, in some embodiments, the present invention can also be implemented in the form of a computer program product in one or more computer-readable media, and the computer-readable medium contains computer-readable program code.

[0134] Any combination of one or more computer-readable media may be used. A computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, device, or device.

[0135] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A battery loss prediction method based on deep learning, characterized in that: include: S1, determining the operating data of the battery to be predicted under different working conditions; S2, decomposing the operating data to obtain modal functions of different frequency components, determining the time domain features and frequency domain features corresponding to the modal functions under each frequency component, integrating all the time domain features and all the frequency domain features to generate a feature vector; S3, processing the feature vector according to the LSTM model to obtain the battery loss result corresponding to the battery to be predicted.

2. The battery loss prediction method based on deep learning according to claim 1, characterized in that: The process of decomposing the operating data to obtain the modal functions of different frequency components is specifically as follows: Decomposing the operating data by using an optimized variational modal decomposition method to obtain modal functions of different frequency components; The optimized variational mode decomposition method is a variational mode decomposition method after parameter optimization, and the parameter optimization process is specifically as follows: Determine multiple training mode functions corresponding to the training operation data, and determine the center frequency corresponding to each training mode function, and by combining constraints, determine that the sum of the multiple training mode functions is equal to the original signal, and use the parameters of the variational mode decomposition method corresponding to the sum of the multiple training mode functions being equal to the original signal as the optimal parameters, and the variational mode decomposition method determined based on the optimal parameters is the optimized variational mode decomposition method.

3. The battery loss prediction method based on deep learning according to claim 1, characterized in that: The LSTM model consists of an input layer, an LSTM layer, a fully connected layer and an output layer; The LSTM layer is composed of multiple neural units, each of which includes a forget gate, an input gate, a memory cell, and an output gate.

4. The battery loss prediction method based on deep learning according to claim 1, characterized in that: Also includes: Preprocessing the operation data to obtain preprocessed operation data, using the preprocessed operation data as the operation data in S2 and executing S2; The preprocessing includes: cleaning and normalizing the operation data.

5. A battery loss prediction system based on deep learning, characterized in that: include: The determination module is used to: determine the operating data of the battery to be predicted under different working conditions; The decomposition module is used to: decompose the operating data to obtain modal functions of different frequency components, determine the time domain features and frequency domain features corresponding to the modal function under each frequency component, integrate all the time domain features and all the frequency domain features, and generate a feature vector; The prediction module is used to: process the feature vector according to the LSTM model to obtain the battery loss result corresponding to the battery to be predicted.

6. A battery loss prediction system based on deep learning according to claim 5, characterized in that: The process of decomposing the operating data to obtain the modal functions of different frequency components is specifically as follows: Decomposing the operating data by using an optimized variational modal decomposition method to obtain modal functions of different frequency components; The optimized variational mode decomposition method is a variational mode decomposition method after parameter optimization, and the parameter optimization process is specifically as follows: Determine multiple training mode functions corresponding to the training operation data, and determine the center frequency corresponding to each training mode function, and by combining constraints, determine that the sum of the multiple training mode functions is equal to the original signal, and use the parameters of the variational mode decomposition method corresponding to the sum of the multiple training mode functions being equal to the original signal as the optimal parameters, and the variational mode decomposition method determined based on the optimal parameters is the optimized variational mode decomposition method.

7. The battery loss prediction system based on deep learning according to claim 5, characterized in that: The LSTM model consists of an input layer, an LSTM layer, a fully connected layer and an output layer; The LSTM layer is composed of multiple neural units, each of which includes a forget gate, an input gate, a memory cell, and an output gate.

8. The battery loss prediction system based on deep learning according to claim 5, characterized in that: Also includes: The preprocessing module is used to: preprocess the operation data to obtain preprocessed operation data, use the preprocessed operation data as the operation data in the decomposition module and run the decomposition module; The preprocessing includes: cleaning and normalizing the operation data.

9. An electronic device, characterized in that: The electronic device comprises a processor, the processor is coupled to a memory, at least one computer program is stored in the memory, and the at least one computer program is loaded and executed by the processor so that the electronic device implements the method according to any one of claims 1 to 4.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores at least one computer program, and the at least one computer program is loaded and executed by a processor so that a computer implements the method according to any one of claims 1 to 4.

Images