Method for predicting remaining service life of transformer substation battery based on electrochemical impedance spectroscopy characteristics

Abstract

The invention relates to a method for predicting the remaining service life of a transformer substation battery based on electrochemical impedance spectroscopy characteristics. The method comprises the following steps: step 1, collecting battery EIS data; step 2, preprocessing and normalizing the collected data; step 3, performing reconstruction denoising on the input data by using DAE, and extracting features; 4, inputting the features extracted by the DAE into a Transform model, capturing the long-time dependency of an input sequence through a multi-head self-attention mechanism, and training the model; 5, evaluating errors of the energy storage battery RUL prediction model, and when the errors of the energy storage battery RUL prediction model are smaller than a preset value, ending prediction; and otherwise, returning to the step 3 for optimization, and repeating the steps 4-5 until the prediction model passes evaluation. The method has high robustness, is suitable for different environment conditions and operation scenes, and significantly improves the prediction accuracy and reliability.

Description

Technical Field

[0001] The present invention belongs to the technical field of battery remaining life prediction, and relates to a method for predicting the remaining service life of a substation battery based on electrochemical impedance spectroscopy characteristics, in particular to a method for predicting the RUL of a substation battery using a DAE-Transformer model based on battery EIS characteristics. Background Art

[0002] With the rapid development of renewable energy storage systems, the widespread use of lithium-ion batteries has made the monitoring and management of battery health status particularly important. The remaining useful life (RUL) of a battery is a key indicator for evaluating battery performance and formulating maintenance strategies. Accurately predicting RUL can help optimize the battery management system (BMS), improve battery efficiency, and reduce safety risks.

[0003] Electrochemical impedance spectroscopy (EIS) is an effective non-destructive testing technology that can provide in-depth information about the internal state of the battery. By analyzing EIS data, the electrochemical reaction kinetics and aging mechanism of the battery can be revealed, thus providing a reliable characteristic basis for RUL prediction. However, traditional model-based RUL prediction methods often rely on complex battery physical models, which are difficult to handle nonlinear characteristics and variable environmental factors, and accurate estimation of model parameters is often challenging.

[0004] In recent years, deep learning methods have shown superior performance in the field of time series forecasting. As an unsupervised learning model, the denoising autoencoder (DAE) can effectively extract features from noisy data and improve the quality and availability of data. Combined with the self-attention mechanism of the Transformer model, the long-term dependencies in the EIS data can be further captured, thereby improving the accuracy of RUL prediction. Summary of the invention

[0005] The purpose of the present invention is to provide a method for predicting the remaining service life of substation batteries based on electrochemical impedance spectroscopy characteristics to address the shortcomings of traditional model-based RUL prediction methods. This method uses DAE to denoise and extract features from EIS data, and uses the Transformer model to capture time series information to achieve accurate prediction of battery RUL, making the prediction results more consistent with the true value. It can not only improve the prediction accuracy, but also provide an important basis for battery status monitoring and management.

[0006] In order to achieve the above object, the present invention is implemented by adopting the following technical solutions:

[0007] A method for predicting the remaining service life of a substation battery based on electrochemical impedance spectroscopy characteristics comprises the following steps:

[0008] Step 1, collecting battery EIS data;

[0009] Step 2, preprocessing and normalizing the collected EIS data;

[0010] Step 3, using a denoising autoencoder (DAE) to reconstruct and denoise the input data and extract features, wherein the input data is the data processed in step 2;

[0011] Step 4: Input the features extracted by the denoising autoencoder into the Transformer model, capture the long-term dependencies of the input sequence through the multi-head self-attention mechanism, and train the model;

[0012] Step 5: perform error evaluation on the energy storage battery RUL prediction model trained in step 4. When the battery RUL prediction model error is less than the preset value, end the prediction; otherwise, return to the above step 3 for optimization, and repeat steps 4 to 5 until the prediction model passes the evaluation.

[0013] Furthermore, the specific method for collecting battery EIS data in step 1 is: using an electrochemical workstation to perform an electrochemical impedance spectroscopy (EIS) test on the battery, collecting data through information such as frequency range, frequency resolution, voltage range and current range. The collected data include parameters such as solution resistance, charge transfer resistance, diffusion impedance or impedance, double layer capacitance, diffusion electrochemical impedance and battery phase angle. These parameters can be obtained by fitting an equivalent circuit model for EIS and are used to evaluate the performance and health status of the battery.

[0014] Furthermore, in step 2, the specific process of preprocessing and normalizing the collected EIS data is as follows:

[0015] Step 2-1: clean the data and divide the data into training set and test set;

[0016] Step 2-2, perform data preprocessing and normalization, use minimum-maximum standardization to perform linear transformation on the sample data, and map the result value to [0,1]; where the sample data is EIS data, the conversion expression is:

[0017] x′=(xx min ) / (x max -x min )

[0018] In the above formula, x is the actual value of the sample data, x′ is the normalized value of the sample data, and x min is the minimum value in the sample data set, x max is the maximum value in the sample data set.

[0019] Furthermore, in step 3, the method of reconstructing, denoising and extracting features is:

[0020] Step 3-1, before compressing the original data, the denoising autoencoder adds Gaussian white noise to the input data to turn it into an input vector, in order to increase the robustness and nonlinear expression ability of the model;

[0021] Step 3-2, the denoising autoencoder compresses the above input vector into a low-dimensional intermediate vector;

[0022] Step 3-3: The decoder reconstructs information and maps the intermediate vector carrying the original data information into an output vector with the same size as the input data.

[0023] Furthermore, in step 3, in order to increase the sparsity of the network and alleviate the overfitting phenomenon, the ReLU function is used as the activation function of the hidden layer in the encoder and decoder; in order to effectively map the output data to the valid value range, the Sigmoid function is used as the activation function of the output layer;

[0024] The loss function of the denoising autoencoder (DAE) part is defined as:

[0025]

[0026] Where n is the number of samples, l(~) is the activation function of the loss, To reconstruct the error, due to the symmetry of the DAE model structure, the weights λ of the encoding matrix W′ and the decoding matrix W are bound to reduce the training time of the model.

[0027] Furthermore, when training the model in step 4, the input features are nonlinearly transformed and feature mapped by the Transformer model to generate new feature representations containing rich time series information, and finally the linear layer performs feature mapping to obtain the predicted sequence results; the specific process includes:

[0028] Step 4-1, a Transformer encoder extracts degradation features from the reconstructed data, wherein the Transformer encoder has a multi-head self-attention layer and a feed-forward neural network layer;

[0029] Step 4-2, use sine and cosine functions of different frequencies to inject relative position markers into the sequence to make full use of the position information of the sequence. The expression is as follows:

[0030] PE(t,2k)=sin(t / 10000 2k / s )

[0031] PE(t,2k+1)=cos(t / 10000 2k / s )

[0032] Where PE represents the position encoding matrix; s represents the dimension of each position vector in the Transformer model; t represents the position of the position encoding; k is an integer value between 0 and (s / 2-1), representing the dimension index of the encoding vector;

[0033] The sequence is the data sequence input into the Transforme model, that is, the data after processing and noise reduction.

[0034] Furthermore, the feedforward network of the Transformer model consists of two linear transformation layers and an activation function, wherein the activation function of the first fully connected layer is a ReLU activation function. It can be expressed as:

[0035] FFN(x)=ReLU(xW1+b1)W2+b2

[0036] Where W1 and W2 are the weights of the two linear layers respectively; b1 and b2 are the biases of the two linear layers.

[0037] Furthermore, in step 5, in order to predict the remaining battery life, a fully connected layer is used to map the last Transformer unit to obtain the final prediction result. The formula is as follows:

[0038]

[0039] Where W p represents the weight of the prediction layer, b p represents the bias of the prediction layer, H h represents the input of the prediction layer, and f(·) represents the mapping function of the prediction layer.

[0040] The present invention also provides a system for predicting the remaining service life of a substation battery based on electrochemical impedance spectroscopy characteristics, comprising:

[0041] Battery EIS data acquisition module, used for battery EIS data acquisition;

[0042] A data preprocessing module is used to preprocess and normalize the battery EIS data sent by the battery EIS data acquisition module;

[0043] A data processing module is used to reconstruct, denoise and extract features from the preprocessed data;

[0044] Model training module, used to train the processed data using the Transformer model to predict the remaining battery life;

[0045] The model evaluation module is used to perform error evaluation on the trained energy storage battery remaining service life prediction model.

[0046] Furthermore, the battery EIS data acquisition module uses an electrochemical workstation to perform an electrochemical impedance spectroscopy (EIS) test on the battery, and collects data through information such as frequency range, frequency resolution, voltage range and current range. The collected data include parameters such as solution resistance, charge transfer resistance, diffusion impedance or impedance, double layer capacitance, diffusion electrochemical impedance and battery phase angle.

[0047] The beneficial effects of the present invention are as follows:

[0048] The DAE-Transformer model based on EIS in the battery management system proposed in this paper effectively denoises and extracts the features of battery EIS data through the denoising autoencoder (DAE), significantly improving the accuracy and reliability of prediction. At the same time, combined with the self-attention mechanism of the Transformer, the model can effectively capture the long-term dependencies in the data, make full use of historical information, and improve the understanding of the battery aging process.

[0049] In addition, this method has strong robustness and can maintain good prediction performance under various battery conditions and adapt to different environmental conditions and operating scenarios. This feature enables the model to be used to monitor battery status in real time, detect potential faults in a timely manner and issue early warnings, thereby effectively improving the safety and reliability of the battery.

[0050] Finally, through accurate RUL prediction, the battery management system can optimize the charging and discharging strategy, reduce operating and maintenance costs, and extend the battery life. This not only achieves efficient use of resources, but also promotes the intelligent development of battery management technology, providing an important reference for future research and application. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The present invention will be further described below in conjunction with the accompanying drawings:

[0052] Figure 1 This is a flow chart of a method for predicting the remaining service life of a substation battery based on electrochemical impedance spectroscopy characteristics according to Example 1 of the present invention.

[0053] Figure 2 Schematic diagram of the operation of the denoising autoencoder in the DAE-Transformer model based on the electrochemical impedance spectroscopy (EIS) characteristics of the battery in Example 2 of the present invention.

[0054] Figure 3 Schematic diagram of the Transformer structure of the DAE-Transformer model based on the electrochemical impedance spectroscopy (EIS) characteristics of the battery in Example 3 of the present invention.

[0055] Figure 4This is a schematic diagram of the structure of a system for predicting the remaining service life of a substation battery based on electrochemical impedance spectroscopy characteristics according to Example 4 of the present invention;

[0056] Figure 5 A schematic diagram of the structure of a device for implementing a method for predicting the remaining useful life of a substation battery based on electrochemical impedance spectroscopy characteristics in Example 5 of the present invention. DETAILED DESCRIPTION

[0057] The exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present invention and to enable the scope of the present invention to be fully communicated to those skilled in the art.

[0058] Embodiment 1:

[0059] like Figure 1 As shown, the present invention provides a method for predicting the remaining service life of a substation battery based on electrochemical impedance spectroscopy characteristics, comprising the following steps:

[0060] Step 1, collecting battery EIS data;

[0061] Step 2, preprocessing and normalizing the collected data;

[0062] Step 3, using a denoising autoencoder (DAE) to reconstruct and denoise the input data and extract features;

[0063] Step 4: Input the features extracted by DAE into the Transformer model, capture the long-term dependencies of the input sequence through the multi-head self-attention mechanism, and train the model;

[0064] Step 5, evaluate the error of the energy storage battery RUL prediction model trained in step 4. When the battery RUL prediction model error is less than the preset value, end the prediction; otherwise, return to the above step 3 for optimization, and repeat steps 4 to 5 until the prediction model passes the evaluation.

[0065] The data normalization process is specifically as follows: linearly transform the sample data using minimum and maximum standardization, and map the result value to between [0,1]. The sample data is battery EIS data, and the conversion expression is:

[0066] x′=(xx min ) / (x max -x min ),

[0067] In the above formula, x and x' are the actual value and normalized value of the sample data respectively, and xmin 、x max are the minimum and maximum values ​​in the sample data set, respectively.

[0068] Embodiment 2:

[0069] In this embodiment, the denoising autoencoder in the DAE-Transformer model based on the electrochemical impedance spectroscopy (EIS) characteristics of the battery uses DAE reconstruction, denoising and feature extraction after the collected EIS data is normalized:

[0070] First, the denoising autoencoder sends the input data x′ to the original data before compressing it. t ={x t+1 ,x t+2 …,x t+n Gaussian white noise is added to} (indicating that the collected data is input in sequence) to turn it into an input vector in order to increase the robustness and nonlinear expression ability of the model.

[0071] Then, the denoising autoencoder compresses the input data into a low-dimensional intermediate vector as follows:

[0072]

[0073] Where W is the denoising autoencoder weight matrix, b is the bias term, a() is the activation function, and z is the compressed hidden vector.

[0074] Finally, the denoising autoencoder reconstructs the information and maps the intermediate vector carrying the original data information into an output vector of the same size as the input data. The formula is as follows:

[0075]

[0076] Where W′ is the weight matrix of the denoising autoencoder; b′ is the bias term; f′(·) is the activation function; is the decoded output vector.

[0077] In order to increase the sparsity of the network and alleviate the overfitting phenomenon, the ReLU function is used as the activation function of the hidden layer in the encoder and decoder. In order to effectively map the output data to the valid value range, the Sigmoid function is used as the activation function of the output layer; the loss function of the denoising autoencoder (DAE) part is defined as:

[0078]

[0079] Where n is the number of samples, l(~) is the activation function of the loss, To reconstruct the error, due to the symmetry of the DAE model structure, the weights λ of the encoding matrix W′ and the decoding matrix W are bound to reduce the training time of the model.

[0080] Embodiment 3:

[0081] Refer to the attached Figure 3 , a schematic diagram of the Transformer structure of the DAE-Transformer model based on the battery electrochemical impedance spectroscopy (EIS) features. In this embodiment, after data reconstruction, denoising and feature extraction are completed, the features extracted by DAE are input into the Transformer model, and the long-term dependencies of the input sequence are captured through the multi-head self-attention mechanism.

[0082] In order to make full use of the position information of the sequence, some relative position markers are injected into the sequence using sine and cosine functions of different frequencies. The expressions are as follows:

[0083] PE(t,2k)=sin(t / 10000 2k / s )

[0084] PE(t,2k+1)=cos(t / 10000 2k / s )

[0085] Where PE represents the position encoding matrix; s represents the dimension of each position vector in the Transformer model; t represents the position of the position encoding; k is an integer value between 0 and (s / 2-1), representing the dimension index of the encoding vector.

[0086] The mathematical expression of multi-head self-attention is as follows:

[0087] MultiHead(Q,K,V)=Concat(head1…head h )W o

[0088] head i =Attention(QW i q ,KW i κ ,VW i v )

[0089] Where Q(query), K(key), and V(value) are three vectors obtained by mapping the input sequence through three linear mapping layers, and their dimensions are d q ,d k ,d v ; Concat represents the concatenation function, which concatenates the output results of all heads.

[0090] They respectively represent the weight matrices corresponding to the Q, K, and V vectors of the i-th head and the weight matrix obtained by dimensionality reduction to obtain the final output.

[0091] In this embodiment, the number of self-attention heads is set to 8, that is, h = 8, and d k =d v =s / h. Scaled dot-product attention is defined as follows:

[0092]

[0093] Where, d h =s / h; softmax is the activation function; K T is the transpose of K. This definition of attention avoids gradient disappearance.

[0094] The feedforward network of Transformer consists of two linear transformation layers and an activation function, where the activation function of the first fully connected layer is the ReLU activation function. It can be expressed as:

[0095] FFN(x)=ReLU(xW1+b1)W2+b2

[0096] Where W1 and W2 are the weights of the two linear layers respectively; b1 and b2 are the biases of the two linear layers.

[0097] In order to predict the remaining battery life, a fully connected layer is used to map the last Transformer unit to obtain the final prediction result. The formula is as follows:

[0098]

[0099] Where W p 、b p , H h and f(·) represent the weight, bias, input, and mapping function of the prediction layer, respectively.

[0100] In order to fully demonstrate the improvement of the prediction accuracy of the model, the experiment of this embodiment uses the LSTM model, the Transformer model and the DAE-Transformer model used in the present invention for training and prediction, and calculates the relative error (RE), root mean square error (RMSE), and mean absolute error (MAE) of the model prediction.

[0101] Table 1 Error comparison of different models

[0102]

[0103] From the calculation results in Table 1, it can be seen that the evaluation indicators RMSE, MAE, and RE show a downward trend from LSTM to DAE-Transformer, and the prediction results are closer to the true values, indicating the effectiveness of DAE in denoising and extracting features from EIS data and the better performance of using the Transformer model to capture time series information to achieve accurate prediction of battery RUL.

[0104] Embodiment 4:

[0105] Refer to the attached Figure 4 The present invention provides a system for predicting the remaining service life of a substation battery based on electrochemical impedance spectroscopy characteristics, comprising:

[0106] Battery EIS data acquisition module, used for battery EIS data acquisition;

[0107] A data preprocessing module is used to preprocess and normalize the battery EIS data sent by the battery EIS data acquisition module;

[0108] A data processing module is used to reconstruct, denoise and extract features from the preprocessed data;

[0109] Model training module, used to train the processed data using the Transformer model to predict the remaining battery life;

[0110] The model evaluation module is used to perform error evaluation on the trained energy storage battery remaining service life prediction model.

[0111] Embodiment 5:

[0112] Refer to the attached Figure 5 The present invention provides a device and apparatus for realizing a method for predicting the remaining service life of a substation battery based on electrochemical impedance spectroscopy characteristics.

[0113] Device 12 is embodied in the form of a general purpose computing device. Components of device 12 may include, but are not limited to: one or more processors or processing units 16, system memory 28, and bus 18 that connects various system components, including system memory 28 and processing unit 16.

[0114] Bus 18 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor or a local bus using any of a variety of bus architectures. These architectures include, but are not limited to, an Industry Standard Architecture (ISA) bus, a Micro Channel Architecture (MAC) bus, an Enhanced ISA bus, a Video Electronics Standards Association (VESA) local bus, and a Peripheral Component Interconnect (PCI) bus.

[0115] Device 12 includes a variety of computer system readable media. These media can be any available media that can be accessed by device 12, including volatile and non-volatile media, removable and non-removable media.

[0116] The system memory 28 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache memory 32. The device 12 may further include other removable / non-removable, volatile / non-volatile computer system storage media. The storage system 34 may be used to read and write non-removable, non-volatile magnetic media, commonly referred to as "hard drives". A magnetic disk drive for reading and writing to removable non-volatile magnetic disks (such as "floppy disks"), and an optical disk drive for reading and writing to removable non-volatile optical disks (such as CD-ROMs, DVD-ROMs or other optical media) may be provided. In these cases, each drive may be connected to the bus 18 via one or more data media interfaces. The memory 28 may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of various embodiments of the present invention.

[0117] A program / utility 40 having a set (at least one) of program modules 42 may be stored, for example, in the memory 28, such program modules 42 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. The program modules 42 generally perform the functions and / or methods of the embodiments described herein.

[0118] The device 12 may also communicate with one or more external devices 14 (e.g., a keyboard, a pointing device, a display 24, etc.), one or more devices that enable a user to interact with the device 12, and / or any device that enables the device 12 to communicate with one or more other computing devices (e.g., a network card, a modem, etc.). Such communication may be performed through an input / output (I / O) interface 22. Furthermore, the device 12 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) through a network adapter 20. Figure 3 As shown, the network adapter 20 communicates with other modules of the device 12 via the bus 18. It should be understood that other hardware and / or software modules may be used in conjunction with the device 12, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, etc.

[0119] The processing unit 16 executes various functional applications and data processing by running programs stored in the system memory 28, such as an implementation program of the method for predicting the remaining service life of a substation battery based on electrochemical impedance spectroscopy characteristics provided in an embodiment of the present invention.

[0120] The present invention also provides a storage medium having a computer program stored thereon, and when the computer program is executed by at least one processor, the steps of the method for predicting the remaining service life of a substation battery based on electrochemical impedance spectroscopy characteristics are implemented.

[0121] The computer storage medium of the embodiment of the present invention can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, a system, device or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (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 of the above. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program, which can be used by an instruction execution system, a device or a device or used in combination with it.

[0122] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0123] The program code embodied on the computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0124] Of course, the storage medium containing computer executable instructions provided by the embodiment of the present invention, whose computer executable instructions are not limited to the above method operations, can also execute related operations in any embodiment of the present invention.

Claims

1. A method for predicting the remaining service life of a substation battery based on electrochemical impedance spectroscopy characteristics, characterized in that: The following steps are involved: Step 1, collecting battery EIS data; Step 2, preprocessing and normalizing the collected EIS data; Step 3, using a denoising autoencoder DAE to reconstruct and denoise the input data and extract features, the input data being the data processed in step 2; Step 4: Input the features extracted by the denoising autoencoder into the Transformer model, capture the long-term dependencies of the input sequence through the multi-head self-attention mechanism, and train the model; Step 5: perform error evaluation on the energy storage battery RUL prediction model trained in step 4. When the battery RUL prediction model error is less than the preset value, end the prediction; otherwise, return to the above step 3 for optimization, and repeat steps 4 to 5 until the prediction model passes the evaluation.

2. The method for predicting the remaining service life of a substation battery based on electrochemical impedance spectroscopy characteristics according to claim 1 is characterized in that: The specific method for collecting battery EIS data in step 1 is: use an electrochemical workstation to perform an electrochemical impedance spectroscopy (EIS) test on the battery, and collect data through frequency range, frequency resolution, voltage range and current range information. The collected data include solution resistance, charge transfer resistance, diffusion impedance or impedance, double layer capacitance, diffusion electrochemical impedance and phase angle parameters of the battery.

3. The method for predicting the remaining service life of a substation battery based on electrochemical impedance spectroscopy characteristics according to claim 1 is characterized in that: In step 2, the specific process of preprocessing and normalizing the collected EIS data is as follows: Step 2-1: clean the data and divide the data into training set and test set; Step 2-2, perform data preprocessing and normalization, use minimum-maximum standardization to perform linear transformation on the sample data, and map the result value to [0,1]; where the sample data is EIS data, the conversion expression is: x′=(x-x min ) / (x max -x min ) In the above formula, x is the actual value of the sample data, x′ is the normalized value of the sample data, and x min is the minimum value in the sample data set, x max is the maximum value in the sample data set.

4. The method for predicting the remaining service life of a substation battery based on electrochemical impedance spectroscopy characteristics according to claim 1 is characterized in that: In step 3, the method of reconstructing, denoising and extracting features is: Step 3-1, before compressing the original data, the denoising autoencoder adds Gaussian white noise to the input data to turn it into an input vector, in order to increase the robustness and nonlinear expression ability of the model; Step 3-2, the denoising autoencoder compresses the above input vector into a low-dimensional intermediate vector; Step 3-3, the denoising autoencoder reconstructs information and maps the intermediate vector carrying the original data information into an output vector with the same size as the input data.

5. The method for predicting the remaining service life of a substation battery based on electrochemical impedance spectroscopy characteristics according to claim 4 is characterized in that: In step 3, in order to increase the sparsity of the network and alleviate the overfitting phenomenon, the ReLU function is used as the activation function of the hidden layer in the encoder and decoder; in order to effectively map the output data to the valid value range, the Sigmoid function is used as the activation function of the output layer; The loss function of the denoising autoencoder (DAE) part is defined as: Where n is the number of samples, l(~) is the activation function of the loss, To reconstruct the error, due to the symmetry of the DAE model structure, the weights λ of the encoding matrix W′ and the decoding matrix W are bound to reduce the training time of the model.

6. The method for predicting the remaining service life of a substation battery based on electrochemical impedance spectroscopy characteristics according to claim 1, characterized in that: When training the model in step 4, the input features are nonlinearly transformed and feature mapped through the Transformer model to generate new feature representations containing rich time series information. Finally, the linear layer performs feature mapping to obtain the predicted sequence results. The specific process includes: Step 4-1, a Transformer encoder extracts degradation features from the reconstructed data, wherein the Transformer encoder has a multi-head self-attention layer and a feed-forward neural network layer; Step 4-2, use sine and cosine functions of different frequencies to inject relative position markers into the sequence to make full use of the position information of the sequence. The expression is as follows: PE(t,2k)=sin(t / 10000 2k / s ) PE(t,2k+1)=cos(t / 10000 2k / s ) Where PE represents the position encoding matrix; s represents the dimension of each position vector in the Transformer model; t represents the position of the position encoding; k is an integer value between 0 and (s / 2-1), representing the dimension index of the encoding vector; The sequence is the data sequence input into the Transforme model, that is, the data after processing and noise reduction.

7. The method for predicting the remaining service life of a substation battery based on electrochemical impedance spectroscopy characteristics according to claim 6 is characterized in that: The feedforward network of the Transformer model consists of two linear transformation layers and an activation function, where the activation function of the first fully connected layer is the ReLU activation function; it is expressed as: FFN(x)=ReLU(xW1+b1)W2+b2 Where W1 and W2 are the weights of the two linear layers respectively; b1 and b2 are the biases of the two linear layers.

8. The method for predicting the remaining service life of a substation battery based on electrochemical impedance spectroscopy characteristics according to claim 1, characterized in that: In step 5, in order to predict the remaining battery life, a fully connected layer is used to map the last Transformer unit to obtain the final prediction result. The formula is as follows: Where W p represents the weight of the prediction layer, b p represents the bias of the prediction layer, H h represents the input of the prediction layer, and f(·) represents the mapping function of the prediction layer.

9. A system for predicting the remaining service life of a substation battery based on electrochemical impedance spectroscopy characteristics for implementing the method described in any one of claims 1 to 8, characterized in that: include: Battery EIS data acquisition module, used for battery EIS data acquisition; A data preprocessing module is used to preprocess and normalize the battery EIS data sent by the battery EIS data acquisition module; A data processing module is used to reconstruct, denoise and extract features from the preprocessed data; Model training module, used to train the processed data using the Transformer model to predict the remaining battery life; The model evaluation module is used to perform error evaluation on the trained energy storage battery remaining service life prediction model.

10. A computer device, characterized in that: The invention comprises a system memory and a processor or a processing unit, wherein the system memory stores a computer program, and the system memory stores a computer program, and the processor or the processing unit implements the method for predicting the remaining service life of a substation battery based on electrochemical impedance spectroscopy characteristics as described in any one of claims 1 to 8 when executing the computer program.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that: The storage medium includes a stored program, wherein when the program is running, the device where the storage medium is located is controlled to execute the method for predicting the remaining service life of a substation battery based on electrochemical impedance spectroscopy characteristics according to any one of claims 1 to 8.

Images