Electrochemical energy storage battery data analysis method and system based on L-S neural network

Abstract

The invention discloses an electrochemical energy storage battery data analysis method and system based on an L-S neural network, and relates to the technical field of intelligent fault diagnosis of an electrochemical energy storage system, and the method comprises the steps: collecting the operation data of an electrochemical energy storage battery, and generating a tag; performing data enhancement to construct a battery fault state monitoring model; and analyzing the battery operation state by using the model to obtain a fault state prediction result. According to the electrochemical energy storage battery data analysis method based on the L-S neural network, data enhancement is performed through linear interpolation, a synthetic sample is generated to enrich a sample space, and the precision and generalization ability of model training are improved; according to the method, the LWDA-AE network is pre-trained layer by layer, encoders of the LWDA-AE network are stacked to construct the LWDA-SAE model, reconstruction errors are optimized, potential relations among data layers are deeply mined, and therefore the fault state monitoring precision is remarkably improved, and better effects are achieved in the aspects of electrochemical energy storage battery monitoring precision, diagnosis efficiency and accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent fault diagnosis of electrochemical energy storage systems, and in particular to a method and system for analyzing electrochemical energy storage battery data based on an LS neural network. Background Art

[0002] In recent years, the proportion of clean energy in the power system has continued to rise, especially renewable energy such as solar energy and wind energy. However, its power supply stability is limited by natural factors such as weather conditions and time periods. Therefore, in order to ensure the overall stability and reliable operation of the power system, large-scale chemical energy storage technology has been rapidly promoted, among which the application scope of electrochemical energy storage batteries is expanding. However, given the high complexity of electrochemical energy storage battery systems, the realization of real-time status monitoring of these batteries and accurate identification of fault points has become a key problem that needs to be solved urgently. Through in-depth data analysis of electrochemical energy storage batteries, it is possible to gain insight into and predict potential faults in advance, ensure the normal operation of energy storage systems, and provide a strong basis for the implementation of preventive maintenance strategies. Furthermore, data analysis can also help operators adjust system operating parameters in a timely manner, effectively prevent system damage, and reduce operating costs. At present, the data analysis methods of electrochemical energy storage batteries cover a variety of approaches such as modeling-based methods and hybrid strategies. However, given the inherent complexity and uncertainty of electrochemical energy storage battery systems, data analysis still faces many challenges. In the future, exploring more efficient data analysis technologies will be particularly important to ensure the safe and reliable operation of electrochemical energy storage battery systems.

[0003] The SAE model is composed of the encoder parts of multiple AEs stacked together. By reconstructing the input at the output layer, SAE can learn the intrinsic features of the input data and use it for subsequent classification, regression and other tasks. Compared with other deep learning algorithms, SAE has the advantages of simple structure, strong feature extraction ability, and easy expansion. In order to solve the problem that insufficient sample data affects the training accuracy of the SAE model, the layer-by-layer data enhancement strategy is used to fully enrich the training samples and improve the generalization ability of the model. Summary of the invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by the present invention is that the existing methods for real-time status monitoring of batteries and accurate identification of fault points have the problems of low identification accuracy, low operating safety, high operating costs, and inability to gain insight into and predict potential faults in advance.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a method for analyzing electrochemical energy storage battery data based on LS neural network, comprising collecting electrochemical energy storage battery operation data to generate labels; performing data enhancement to build a battery fault state monitoring model; and using the model to analyze the battery operation state to obtain a fault state prediction result.

[0007] As a preferred solution of the LS neural network electrochemical energy storage battery data analysis method of the present invention, wherein: the collection of electrochemical energy storage battery operation data to generate labels includes preprocessing the historical operation data recorded by the electrochemical energy storage battery monitoring system, including battery voltage, current, state of charge and health status information;

[0008] Eliminate incomplete and inaccurate data and extract key features of the data.

[0009] As a preferred embodiment of the LS neural network-based electrochemical energy storage battery data analysis method of the present invention, the collection of electrochemical energy storage battery operation data to generate labels includes combining historical fault records, maintenance files and technical guidelines of the electrochemical energy storage battery, analyzing expert knowledge of electrochemical energy storage battery system operation risk assessment and maintenance strategies, and constructing an expert experience model based on this.

[0010] As a preferred solution of the LS neural network electrochemical energy storage battery data analysis method of the present invention, wherein: the data enhancement to build a battery fault state monitoring model includes performing data enhancement on the original training samples to generate synthetic samples, and the synthetic samples x v The two adjacent input vectors x in the original training data i and x j The linear interpolation method is used to generate the expression:

[0011] x v =(x i +x j ) / 2

[0012] Among them, x v represents a synthetic sample, x i and x j represents the input vector;

[0013] The data augmentation strategy applied to all hidden layers of the neural network is expressed as:

[0014]

[0015] in, and represents the feature data of the kth hidden layer, represents a synthetic sample;

[0016] Enrich the sample data by linearly interpolating the input layer and hidden layer of the neural network.

[0017] As a preferred solution of the LS neural network-based electrochemical energy storage battery data analysis method described in the present invention, the data enhancement to build a battery fault status monitoring model includes building an LWDA-SAE neural network model, determining the structure and number of nodes of the neural network according to the type of fault data to be diagnosed, initializing weights and bias network parameters, and setting a learning rate and activation function.

[0018] As a preferred solution of the LS neural network electrochemical energy storage battery data analysis method of the present invention, wherein: the use of the model to analyze the battery operating state to obtain the fault state prediction result includes pre-training the LWDA-SAE neural network model, converting the original sample X r ={x r(1) ,x r(2) ,…x r(i) ,…x r(N)} and synthetic sample X v1 ={x v(1) ,x v(2) ,…x v(i) ,...x v(N-1)} are used as the input of the LWDA-AE1 neural network model, and the pre-training is expressed as:

[0019]

[0020] Where N represents the number of original samples, W and b represent the weight matrix and bias vector of the encoder respectively. and Represent the weight matrix and bias vector of the decoder respectively, and Represent the reconstructed original samples and the reconstructed synthetic samples respectively;

[0021] After training, the weight matrix of the encoder part of LWDA-AE1 is recorded as W1 (1) , input data {X r ,X v1}Forward propagation from the input layer to the first hidden layer, output the first layer of hidden features The superscript (k) is used to denote the kth hidden layer;

[0022] For LWDA-AE2, the original sample is the first layer of hidden features Perform data augmentation to obtain synthetic samples By using the training samples Minimize the reconstruction error to pre-train LWDA-AE2;

[0023] After training, the weight of the LWDA-AE2 encoder part is W1 (2) , The second layer of hidden features is obtained through forward propagation

[0024] Alternately perform data augmentation and model pre-training, and output the final LWDA-AE;

[0025] The encoder part of the stacked LWDA-AE outputs LWDA-SAE;

[0026] The quality variables are added to the last layer of LWDA-SAE to build the battery fault status monitoring model LWDA-SAE-NN network, which is used to monitor the battery fault status after fine-tuning the network parameters.

[0027] As a preferred solution of the LS neural network electrochemical energy storage battery data analysis method of the present invention, wherein: the use of the model to analyze the battery operating state to obtain the fault state prediction result includes using the battery fault state monitoring model to perform data analysis on the current battery operating state and output the prediction result of the electrochemical energy storage battery;

[0028] Comprehensively evaluate the operating status of the battery based on the established expert experience model that associates the operating data and status of the electrochemical energy storage battery;

[0029] Based on the evaluation results, the fault location in the electrochemical energy storage battery system is identified and located.

[0030] Another object of the present invention is to provide an electrochemical energy storage battery data analysis system based on an LS neural network, which can construct a battery fault status monitoring model by performing data enhancement, thereby solving the problem of low recognition accuracy in current real-time battery status monitoring and accurate fault point identification methods.

[0031] As a preferred solution of the LS neural network electrochemical energy storage battery data analysis system described in the present invention, it includes: a data preprocessing module, a data enhancement and model pretraining module, and a state monitoring and result prediction module; the data preprocessing module is used to collect and preprocess the operation data of the electrochemical energy storage battery, build a model based on expert experience, associate the battery operation data with the battery state, and generate a battery state label; the data enhancement and model pretraining module is used to perform data enhancement on the original training samples, generate synthetic samples, combine the original samples and the synthetic samples as input, pretrain the LWDA-AE neural network model, and by alternating data enhancement and model pretraining, gradually stack multiple LWDA-AE encoder parts to build a battery fault state monitoring model; the state monitoring and result prediction module is used to use the LWDA-SAE model to analyze the battery operation state and obtain the fault state prediction result of the electrochemical energy storage battery.

[0032] A computer device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement a step of an electrochemical energy storage battery data analysis method based on an LS neural network.

[0033] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of a method for analyzing electrochemical energy storage battery data based on an LS neural network.

[0034] Beneficial effects of the present invention: The LS neural network-based electrochemical energy storage battery data analysis method provided by the present invention effectively eliminates incomplete and inaccurate data by preprocessing the battery operation data, and extracts key features to ensure the reliability and accuracy of subsequent analysis, laying a solid foundation for model construction. Combined with historical fault records and expert knowledge, an expert experience model that associates operation data with status is established to provide a theoretical basis for accurate battery status evaluation and overcome the limitations of a single parameter evaluation method. In the model construction stage, linear interpolation is used to enhance data, generate synthetic samples to enrich the sample space, and improve the accuracy and generalization ability of model training; the LWDA-AE network is pre-trained layer by layer, and its encoder is stacked to construct the LWDA-SAE model, optimize the reconstruction error, and deeply explore the potential relationship between data levels, thereby significantly improving the accuracy of fault status monitoring. In the application stage, the trained model is used to perform real-time analysis and prediction of the battery operation status, significantly enhancing the reliability of fault trend identification. Based on the evaluation results and expert model, accurate fault location and efficient processing are achieved. The overall method achieves high efficiency and accuracy in battery status monitoring through data preprocessing, layer-by-layer data enhancement and model optimization. The present invention achieves better results in electrochemical energy storage battery monitoring precision, diagnostic efficiency and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0036] Figure 1 An overall flow chart of a method for analyzing electrochemical energy storage battery data based on an LS neural network is provided for the first embodiment of the present invention.

[0037] Figure 2 A LWDA-SAE neural network structure diagram of an LS neural network electrochemical energy storage battery data analysis method provided in the first embodiment of the present invention.

[0038] Figure 3 An overall flow chart of an electrochemical energy storage battery data analysis system based on an LS neural network is provided for the third embodiment of the present invention. DETAILED DESCRIPTION

[0039] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0040] Example 1, reference Figure 1 , as an embodiment of the present invention, provides a method for analyzing electrochemical energy storage battery data based on LS neural network, comprising:

[0041] S1: Collect electrochemical energy storage battery operation data to generate labels.

[0042] Furthermore, a large amount of historical operating data recorded by the electrochemical energy storage battery monitoring system, including battery voltage, current, state of charge (SOC) and state of health (SOH) information, is preprocessed. This process covers the elimination of incomplete and inaccurate data and the extraction of key features of the data to ensure the accuracy and effectiveness of subsequent analysis.

[0043] It should be noted that, by combining the historical fault records, maintenance files and technical guidelines of electrochemical energy storage batteries, the expert knowledge of electrochemical energy storage battery system operation risk assessment and maintenance strategy was comprehensively analyzed, and an expert experience model was constructed based on this. This model can map the operating data of the energy storage battery and its actual status, providing strong support for the accurate assessment of the battery status.

[0044] S2: Perform data enhancement to build a battery fault status monitoring model.

[0045] Furthermore, the original training samples are augmented to generate synthetic samples, the synthetic samples x v The two adjacent input vectors x in the original training data i and x j Generate using linear interpolation:

[0046] x v =(x i +x j ) / 2

[0047] Similarly, this data augmentation strategy is applied to all hidden layers of the neural network:

[0048]

[0049] in, and represents the feature data of the kth hidden layer, Represents the corresponding synthetic sample. By performing linear interpolation on the input layer and hidden layer of the neural network, the sample data is enriched, the problem of insufficient samples affecting the accuracy of model training is solved, and the generalization ability of the model is improved.

[0050] It should be noted that constructing the LWDA-SAE neural network model mainly includes: determining the structure and number of nodes of the neural network according to the type of fault data to be diagnosed (the voltage, current, SOC value and SOH value of the electrochemical energy storage battery are mainly monitored in the present invention), initializing network parameters such as weights and biases, and setting the learning rate and activation function.

[0051] S3: Use the model to analyze the battery operating status to obtain fault status prediction results.

[0052] Furthermore, the LWDA-SAE neural network model is pre-trained and the original sample X r ={x r(1 ),x r(2 ),...x r(i ),...x r(N)} and synthetic sample X v1 ={x v(1) ,x v(2),...x v(i ),...x v(N-1)} are used as the input of the LWDA-AE1 neural network model and pre-trained by minimizing the reconstruction error:

[0053]

[0054] Where N represents the number of original samples, W and b represent the weight matrix and bias vector of the encoder respectively. and Represent the weight matrix and bias vector of the decoder respectively, and They represent the reconstructed original samples and the reconstructed synthetic samples respectively.

[0055] After training, the weight matrix of the encoder part of LWDA-AE1 is recorded as W1 (1) , input data {X r ,X v1}Forward propagation from the input layer to the first hidden layer to obtain the first layer of hidden features The superscript (k) is used to denote the kth hidden layer.

[0056] For LWDA-AE2, the original sample is the first layer of hidden features Perform data augmentation to obtain synthetic samples By using the training samples LWDA-AE2 is pre-trained by minimizing the reconstruction error. After training, the weight of the LWDA-AE2 encoder part is W1 (2) , The second layer of hidden features is obtained through forward propagation

[0057] Similarly, data augmentation and model pre-training are performed alternately until the final LWDA-AE is obtained.

[0058] Next, the encoder parts of multiple LWDA-AEs are stacked to obtain LWDA-SAE. The quality variable is added to the last layer of LWDA-SAE to construct the battery fault status monitoring model LWDA-SAE-NN network, which is used to monitor the battery fault status after fine-tuning the network parameters.

[0059] The accuracy rate is used to evaluate the accuracy of electrochemical energy storage battery fault classification. The specific calculation formula is as follows:

[0060]

[0061] Among them, TP represents the number of positive samples predicted by the model as positive, FP represents the number of negative samples predicted by the model as positive, FN represents the number of positive samples predicted by the model as negative, and TN represents the number of negative samples predicted by the model as negative.

[0062] It should be noted that the trained battery fault status monitoring model is used to perform data analysis on the current battery operating status to obtain the prediction result of the electrochemical energy storage battery.

[0063] Furthermore, based on the established expert experience model that associates the operation data and status of electrochemical energy storage batteries, a comprehensive evaluation of the battery operation status is conducted;

[0064] It should be noted that based on the evaluation results, the fault location in the electrochemical energy storage battery system is accurately identified and located.

[0065] The present invention proposes a data analysis model based on LWDA-SAE neural network that can perform real-time monitoring and rapid feedback on electrochemical energy storage batteries.

[0066] The proportion of clean energy in the power system continues to grow, but its power supply stability is restricted by natural factors. In order to ensure the stable operation of the power system, large-scale chemical energy storage technology has developed rapidly, especially electrochemical energy storage batteries have been widely used. However, the high complexity of the electrochemical energy storage battery system makes real-time status monitoring and accurate fault identification a problem that needs to be overcome urgently. In view of the problem that insufficient sample data affects the accuracy of model training in the data analysis of electrochemical energy storage batteries, the present invention fully and effectively expands the sample data by alternately executing data enhancement strategies and model pre-training. While solving the problem of insufficient sample data, real-time monitoring and accurate fault diagnosis of electrochemical energy storage batteries are achieved to ensure the stable operation of the power system.

[0067] Example 2, an embodiment of the present invention, provides a method for analyzing electrochemical energy storage battery data based on LS neural network. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0068] First of all, this experiment aims to verify the performance of the electrochemical energy storage battery data analysis method based on LWDA-SAE neural network in practical applications. The experimental data comes from a certain energy storage battery monitoring system, and the collected historical data includes battery voltage, current, state of charge (SOC) and state of health (SOH). First, the historical operation data was preprocessed to eliminate incomplete records (accounting for 7.5% of the total data) and obvious abnormal points (such as voltage exceeding 4.2V or below 2.5V). At the same time, key features are extracted through principal component analysis to ensure data clarity and accuracy.

[0069] Subsequently, an expert experience model was established by combining the battery's historical failure records and maintenance files with the technical guidelines provided by the battery manufacturer. The model associates the above parameters with the battery's operating status and generates training labels by annotating the operating status, covering three states: normal, slightly aged, and potential failure.

[0070] The data enhancement process uses a linear interpolation method to enhance the original samples to generate synthetic samples, expanding the number of training samples. The training process of the LWDA-SAE model is achieved by stacking encoders layer by layer, optimizing the reconstruction error to improve the model accuracy. Finally, the trained LWDA-SAE neural network is used to monitor the status of test samples in real time and predict the probability of failure.

[0071] Table 1 Experimental data table

[0072]

[0073] It can be seen from the table data that the method of the present invention can significantly reduce the reconstruction error under different battery parameters and effectively predict the probability of potential failure. For example, in sample 3, the SOC is 78% and the SOH reaches 96%, but the model concludes that the probability of failure is 5% through analysis, which reminds that timely inspection is required. Compared with the traditional method that relies on fixed threshold judgment, this method can capture potential problems more sensitively by integrating multiple parameters and expert experience models.

[0074] In addition, the reconstruction error of the model always remains between 0.01 and 0.03, indicating that the data enhancement and layer-by-layer training strategies effectively improve the generalization ability of the model. This optimization solves the problem of insufficient accuracy of traditional technologies due to insufficient data. Compared with existing technologies, this method enriches the sample space through data enhancement and optimizes the network structure through layer-by-layer stacking, which not only improves the prediction accuracy, but also realizes the precision of fault location, showing strong innovation and practicality.

[0075] Example 3, reference Figure 3 , which is an embodiment of the present invention, provides an electrochemical energy storage battery data analysis system based on LS neural network, including a data preprocessing module, a data enhancement and model pre-training module, and a state monitoring and result prediction module.

[0076] The data preprocessing module is used to collect and preprocess the operating data of the electrochemical energy storage battery, build a model based on expert experience, associate the battery operating data with the battery status, and generate a battery status label. The data enhancement and model pretraining module is used to perform data enhancement on the original training samples, generate synthetic samples, combine the original samples and the synthetic samples as input, and pretrain the LWDA-AE neural network model. By alternating data enhancement and model pretraining, multiple LWDA-AE encoder parts are gradually stacked to build a battery fault state monitoring model. The state monitoring and result prediction module is used to use the LWDA-SAE model to analyze the battery operating status and obtain the fault state prediction result of the electrochemical energy storage battery.

[0077] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0078] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0079] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0080] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc. It should be noted that the above embodiments are only used to illustrate the technical solution of the present invention and are not limited. Although the present invention is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

[0081] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for analyzing electrochemical energy storage battery data based on LS neural network, characterized in that: include: Collecting electrochemical energy storage battery operation data to generate labels; Perform data enhancement to build a battery fault status monitoring model; The model is used to analyze the battery operating status and obtain the fault status prediction results.

2. The method for analyzing electrochemical energy storage battery data based on LS neural network according to claim 1, characterized in that: The collecting of electrochemical energy storage battery operation data to generate a label includes preprocessing the historical operation data recorded by the electrochemical energy storage battery monitoring system, including battery voltage, current, state of charge and health status information; Eliminate incomplete and inaccurate data and extract key features of the data.

3. The method for analyzing electrochemical energy storage battery data based on LS neural network as claimed in claim 2, characterized in that: The collection of electrochemical energy storage battery operation data to generate labels includes combining historical fault records, maintenance files and technical guidelines of the electrochemical energy storage battery, analyzing expert knowledge of electrochemical energy storage battery system operation risk assessment and maintenance strategy, and constructing an expert experience model based on this.

4. The method for analyzing electrochemical energy storage battery data based on LS neural network as claimed in claim 3, characterized in that: The data enhancement to construct the battery fault status monitoring model includes performing data enhancement on the original training samples to generate synthetic samples, the synthetic samples x v The two adjacent input vectors x in the original training data i and x j The linear interpolation method is used to generate the expression: x v =(x i +x j ) / 2 Among them, x v represents a synthetic sample, x i and x j represents the input vector; The data augmentation strategy applied to all hidden layers of the neural network is expressed as: in, and represents the feature data of the kth hidden layer, represents a synthetic sample; Enrich the sample data by linearly interpolating the input layer and hidden layer of the neural network.

5. The method for analyzing electrochemical energy storage battery data based on LS neural network according to claim 4, characterized in that: The data enhancement to construct a battery fault status monitoring model includes constructing an LWDA-SAE neural network model, determining the structure and number of nodes of the neural network according to the type of fault data to be diagnosed, initializing weight and bias network parameters, and setting a learning rate and activation function.

6. The method for analyzing electrochemical energy storage battery data based on LS neural network according to claim 5, characterized in that: The method of using the model to analyze the battery operation status to obtain the fault status prediction result includes pre-training the LWDA-SAE neural network model, converting the original sample X r ={x r(1) ,x r(2) ,…x r(i) ,…x r(N) } and synthetic sample X v1 ={x v(1) ,x v(2) ,…x v(i) ,…x v(N-1) } are used as the input of the LWDA-AE1 neural network model, and the pre-training is expressed as: Where N represents the number of original samples, W and b represent the weight matrix and bias vector of the encoder respectively. and Represent the weight matrix and bias vector of the decoder respectively, and Represent the reconstructed original samples and the reconstructed synthetic samples respectively; After training, the weight matrix of the encoder part of LWDA-AE1 is recorded as W1 (1) , input data {X r ,X v1 }Forward propagation from the input layer to the first hidden layer, output the first layer of hidden features The superscript (k) is used to denote the kth hidden layer; For LWDA-AE2, the original sample is the first layer of hidden features Perform data augmentation to obtain synthetic samples By using the training samples Minimize the reconstruction error to pre-train LWDA-AE2; After training, the weight of the LWDA-AE2 encoder part is W1 (2) , The second layer of hidden features is obtained through forward propagation Alternately perform data augmentation and model pre-training, and output the final LWDA-AE; The encoder part of the stacked LWDA-AE outputs LWDA-SAE; The quality variables are added to the last layer of LWDA-SAE to build the battery fault status monitoring model LWDA-SAE-NN network, which is used to monitor the battery fault status after fine-tuning the network parameters.

7. The method for analyzing electrochemical energy storage battery data based on LS neural network according to claim 6, characterized in that: The method of using the model to analyze the battery operating state to obtain the fault state prediction result includes using the battery fault state monitoring model to perform data analysis on the current battery operating state and output the prediction result of the electrochemical energy storage battery; Comprehensively evaluate the operating status of the battery based on the established expert experience model that associates the operating data and status of the electrochemical energy storage battery; Based on the evaluation results, the fault location in the electrochemical energy storage battery system is identified and located.

8. A system using the LS neural network electrochemical energy storage battery data analysis method according to any one of claims 1 to 7, characterized in that: Including data preprocessing module, data enhancement and model pre-training module, status monitoring and result prediction module; The data preprocessing module is used to collect and preprocess the electrochemical energy storage battery operation data, build a model based on expert experience, associate the battery operation data with the battery status, and generate a battery status label; The data enhancement and model pre-training module is used to perform data enhancement on the original training samples, generate synthetic samples, combine the original samples and the synthetic samples as input, pre-train the LWDA-AE neural network model, and gradually stack multiple LWDA-AE encoder parts by alternately performing data enhancement and model pre-training to build a battery fault status monitoring model; The state monitoring and result prediction module is used to analyze the battery operation state using the LWDA-SAE model to obtain the fault state prediction result of the electrochemical energy storage battery.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the LS neural network-based electrochemical energy storage battery data analysis method described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the LS neural network-based electrochemical energy storage battery data analysis method described in any one of claims 1 to 7 are implemented.

Images