Power distribution network grounding fault classification method and system, and medium

Abstract

The invention relates to a power distribution network ground fault classification method and system and a medium, and the method comprises the following steps: collecting various power distribution network ground fault original data, and carrying out the preprocessing of the data; performing label coding on the data, and marking whether each data point belongs to a fault and a fault type; singular value decomposition is carried out on the fault matrix through an improved sparrow search algorithm, and main signal components are extracted; taking the singular value feature vector extracted from the ISSA as the input data of the LSTM model; establishing an LSTM model, and determining model parameters; setting corresponding parameters of the ISSA algorithm and initial hyper-parameters of the LSTM; using ISSA to optimize hyper-parameters in the LSTM network, and combining to form an ISSA-LSTM model; and verifying the ISSA-LSTM model by using the data of the test set. According to the invention, the improved sparrow search algorithm is introduced to optimize the long-short term memory network, and improvement is carried out for the deficiency of power distribution network ground fault classification. The objective of the method is to improve fault classification accuracy.

Description

Technical Field

[0001] The present application relates to the field of distribution network grounding fault classification, and more specifically, to a distribution network grounding fault classification method, system and medium. Background Art

[0002] Each ground fault has its own unique physical mechanism and manifestation, but in actual situations, these fault characteristics are often masked or confused by the complex field environment. The fault signal may contain a lot of noise and interference, making it difficult to extract and identify useful fault features.

[0003] Ground faults include many factors, from natural environmental factors (such as bird nests on wires, birds hitting wires), to human factors (such as cars hitting wires, foreign objects hitting wires), to equipment problems (such as poor transformer insulation, lightning arrester failure), etc. This diversity leads to blurred boundaries between faults, increasing the difficulty of classification.

[0004] In actual operation, some types of ground faults may be more common, while other types of faults are relatively rare. This problem of unbalanced data volume may cause the machine learning algorithm to be biased towards the majority class during training, thereby affecting the ability to identify minority class faults.

[0005] Ground faults need to be identified and handled promptly and accurately to avoid greater impact on power systems and users. However, in practical applications, the real-time and accuracy of algorithms are often limited by multiple factors such as computing resources and data quality.

[0006] Due to the diversity and complexity of ground faults, it is very difficult to design a general algorithm that can be applied to all situations. The algorithm needs to maintain stable performance in different environments, different equipment, and different fault conditions.

[0007] In the field of classification and diagnosis of grounding faults in distribution networks, existing technologies mainly rely on traditional signal processing methods and machine learning algorithms, such as support vector machines (SVM), K-nearest neighbors (KNN) and decision trees (DT). Although these methods can achieve fault classification to a certain extent, they still have the following significant disadvantages: Limited feature extraction capability: Traditional methods often rely on manually designed features, which may not fully reflect the fault characteristics in the complex and changeable distribution network grounding fault environment, resulting in limited classification accuracy. Insufficient ability to process high-dimensional features: With the expansion of the scale of distribution networks and the improvement of their intelligence, fault data is becoming increasingly complex. Traditional methods are unable to cope with high-dimensional features and are difficult to capture subtle differences in fault characteristics. Poor real-time performance: When processing large-scale distribution network data, the computational complexity of traditional methods is high and it is difficult to meet real-time requirements, which is particularly unfavorable for grounding fault classification scenarios that require rapid response. Poor adaptability: The operating environment of the distribution network is complex and changeable, and there are many types of faults. When facing new faults or complex fault scenarios, traditional methods often need to redesign features or adjust parameters, and lack good adaptive capabilities. Difficulty in identifying high-resistance grounding faults: High-resistance grounding faults are difficult to identify effectively using traditional methods due to their large transition resistance and weak fault characteristics, making the handling of such faults a technical challenge. Summary of the invention

[0008] The purpose of the embodiments of the present application is to provide a distribution network grounding fault classification method, system and medium, which combines intelligent optimization algorithms with deep learning technology, provides new ideas and technical means for the intelligent operation and maintenance and fault management of distribution networks, and helps to improve the overall operating efficiency and reliability of the power system.

[0009] To achieve the above objectives, this application provides the following technical solutions:

[0010] In a first aspect, an embodiment of the present application provides a method for classifying ground faults in a distribution network, comprising the following steps:

[0011] Collect various raw data of distribution network grounding faults and pre-process the data;

[0012] Label the data and mark whether each data point is a fault and its fault type;

[0013] The ground fault data is embedded into a high-dimensional trajectory matrix through an improved sparrow search algorithm, and the fault matrix is ​​subjected to singular value decomposition to extract the main signal components.

[0014] The singular value feature vector extracted from ISSA is used as the input data of the LSTM model;

[0015] Establish an LSTM model and determine the parameters of the forget gate, input gate, and output gate in the core part of the model;

[0016] Set the corresponding parameters of the ISSA algorithm and the initial hyperparameters of the LSTM;

[0017] Use ISSA to optimize the hyperparameters in the LSTM network and combine them to form an ISSA-LSTM model;

[0018] The ISSA-LSTM model is validated using the data from the test set.

[0019] The method of collecting a variety of distribution network grounding fault raw data and preprocessing the data is as follows:

[0020] The original data of 15 types of ground faults in distribution networks were collected and preprocessed, including processing missing values, removing noise, smoothing data, and standardizing data to ensure data consistency and accuracy. The missing values ​​of the data were processed using the mean interpolation method. have:

[0021] ,

[0022] in For missing values, use Make a substitution,

[0023] Use the standard deviation method to remove noise and use the mean and standard deviation Determine whether a data point is an outlier. If the data point The value exceeds The range is considered as noise, that is:

[0024]

[0025] Smooth the time series data by exponential weighted average. The simple exponential smoothing formula is:

[0026]

[0027] in is the smoothed value, is the smoothing coefficient ,

[0028] Finally, the processed data is standardized and the eigenvalues ​​are scaled to In scope:

[0029] .

[0030] The data is labeled and encoded to mark whether each data point belongs to a fault and its fault type. Specifically, the fault is set to 15 types of labels from 0 to 14, corresponding to tree branches, tree obstacles, transformer joints, fuse failure, pole switch failure, PT failure, pole underground cable failure, wire burden, poor insulation of ground cable, ground cable discharge, cabinet cable discharge, animal intrusion into the cabinet, low voltage fault, lightning strike, and lightning conductor detachment.

[0031] The improved sparrow search algorithm is used to embed the ground fault data into a high-dimensional trajectory matrix, perform singular value decomposition on the fault matrix, and extract the main signal components, specifically:

[0032] Each fault signal data is constructed into a matrix. Assuming that the number of data points of each signal is N and the length of the signal is L, the signal data can be reconstructed into an L×K trajectory matrix X, where ,

[0033] Perform singular value decomposition on the reconstructed trajectory matrix X. The specific formula is:

[0034]

[0035] Where: is a left singular matrix, is a right singular matrix,

[0036] From the singular value matrix The first few main singular values ​​are selected, which correspond to the main features of the fault signal. By analyzing the size of these singular values, different fault types can be effectively distinguished, thereby achieving the purpose of optimizing the classification ability of LSTM.

[0037] The singular value feature vector extracted from ISSA is used as the input data of the LSTM model as follows:

[0038] The singular value eigenvectors extracted from the ISSA As input data of the LSTM model, for each time step, the LSTM model receives a set of singular value feature vectors, forming a feature sequence input:

[0039] (6)

[0040] In the formula for different time steps.

[0041] The LSTM model is established, and the parameters of the forget gate, input gate, and output gate of the core part of the model are determined as follows: in the model parameter setting, the number of hidden units of the LSTM layer is set to 50, the learning rate is set to 0.001, and the cross entropy function is selected as the loss function, and the activation function adopts the tanh function. During the model training process, iterative optimization is performed through the training set, and the validation set is used for tuning to prevent overfitting. Finally, the trained LSTM model is used to classify the grounding faults of the new data, and the probability distribution of each category is output to determine the specific fault type.

[0042] The parameters corresponding to the ISSA algorithm and the LSTM initial hyperparameters are specifically set as follows:

[0043] The parameters of the ISSA-LSTM model are set, the newly collected unlabeled fault data is input into the pre-trained ISSA-LSTM model, the model is trained repeatedly, and the accuracy of the model in fault classification is verified to form the final ISSA-LSTM model.

[0044] The method of using ISSA to optimize the hyperparameters in the LSTM network and combining them to form the ISSA-LSTM model specifically includes: using ISSA to optimize the hyperparameters of the LSTM, firstly, generating an initial population, each individual representing a set of LSTM hyperparameters, then training the LSTM model and evaluating the performance of each set of parameters, and updating the population according to the fitness evaluation results, generating new individuals, and selecting individuals with higher fitness for the next round of iterations, and finally, iteratively executing the above steps until the maximum number of iterations is reached or the stopping condition is met.

[0045] In a second aspect, an embodiment of the present application provides a distribution network grounding fault classification system, the system comprising: a memory and a processor, the memory comprising a program of a distribution network grounding fault classification method, and the program of the distribution network grounding fault classification method when executed by the processor implements the following steps: collecting a variety of distribution network grounding fault raw data and preprocessing the data; label encoding the data, marking whether each data point belongs to a fault and its fault type; embedding the grounding fault data into a high-dimensional trajectory matrix through an improved sparrow search algorithm, performing singular value decomposition on the fault matrix, and extracting the main signal components; using the singular value eigenvector extracted from ISSA as input data of the LSTM model; establishing an LSTM model, and determining the parameters of the forgetting gate, input gate, and output gate of the core part of the model; setting the corresponding parameters of the ISSA algorithm and the initial hyperparameters of the LSTM; using ISSA to optimize the hyperparameters in the LSTM network, and combining to form an ISSA-LSTM model; and verifying the ISSA-LSTM model using data from a test set.

[0046] In a third aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a program code, and when the program code is executed by a processor, the steps of the distribution network grounding fault classification method as described above are implemented.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] 1. Improve fault classification accuracy:

[0049] (1) Optimizing model performance: By using the improved sparrow search algorithm (ISSA) to optimize the hyperparameters of the LSTM network, the model's ability to learn distribution network fault characteristics can be improved, thereby improving the accuracy of fault classification.

[0050] (2) Capturing complex features: The LSTM network is good at processing time series data and can effectively identify the timing characteristics of fault signals in the distribution network. Combined with ISSA, it can further enhance the recognition of complex fault modes.

[0051] 2. Enhance real-time monitoring capabilities

[0052] (1) Rapid response to faults: Faults in the distribution network need to be quickly located and classified to reduce power outage time and economic losses. ISSA-LSTM can achieve efficient real-time monitoring and analysis, providing timely fault diagnosis information for operation and maintenance personnel.

[0053] (2) Dynamic adjustment: The model can dynamically adjust parameters according to real-time data to adapt to changes in the operating status of the distribution network and improve the flexibility of online fault classification.

[0054] 3. Handling various fault types

[0055] (1) Wide applicability: Various types of faults may occur in the distribution network. ISSA-LSTM can combine the characteristics of different faults for comprehensive learning, thereby improving the ability to classify various types of faults.

[0056] (2) Multidimensional data fusion: The model can integrate data from different sensors and monitoring devices, comprehensively analyze fault conditions, and improve the comprehensiveness and accuracy of classification.

[0057] 4. Reduce false positive rate

[0058] (1) Reduce misdiagnosis: Traditional fault classification methods may result in a high false alarm rate. ISSA-LSTM can improve the model's discrimination ability and reduce the risk of false alarms and missed alarms by optimizing the learning process.

[0059] (2) Intelligent decision support: Higher classification accuracy makes operation and maintenance decisions more reliable and helps to formulate effective maintenance and emergency response strategies.

[0060] 5. Improve training efficiency

[0061] Accelerate model training: ISSA can speed up the training process of the LSTM model. By optimizing algorithm parameters, the model converges faster, improving overall training efficiency and providing a faster solution for practical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments of the present application will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0063] Figure 1 It is a flow chart of the method of this application.

[0064] Figure 2 This is the LSTM model structure diagram of this application. DETAILED DESCRIPTION

[0065] The technical solutions in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application. It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings.

[0066] The terms "comprises," "comprising," or any other variation thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes the element.

[0067] The terms "first", "second", etc. are only used to distinguish one entity or operation from another entity or operation, and should not be understood as indicating or implying relative importance, nor should they be understood as requiring or implying any such actual relationship or order between these entities or operations.

[0068] like Figure 1 The present invention provides a method for classifying ground faults in a distribution network, comprising the following steps:

[0069] Collect various raw data of distribution network grounding faults and pre-process the data;

[0070] Label the data and mark whether each data point is a fault and its fault type;

[0071] The ground fault data is embedded into a high-dimensional trajectory matrix through an improved sparrow search algorithm, and the fault matrix is ​​subjected to singular value decomposition to extract the main signal components.

[0072] The singular value feature vector extracted from ISSA is used as the input data of the LSTM model;

[0073] Establish an LSTM model and determine the parameters of the forget gate, input gate, and output gate in the core part of the model;

[0074] Set the corresponding parameters of the ISSA algorithm and the initial hyperparameters of the LSTM;

[0075] Use ISSA to optimize the hyperparameters in the LSTM network and combine them to form an ISSA-LSTM model;

[0076] The ISSA-LSTM model is validated using the data from the test set.

[0077] Example:

[0078] (1) Collect raw data of various distribution network grounding faults and preprocess the data, including processing missing values, removing noise, smoothing data, and standardizing data to ensure data consistency and accuracy.

[0079] The original data of 15 types of ground faults in distribution networks were collected and preprocessed, including processing missing values, removing noise, smoothing data, and standardizing data to ensure data consistency and accuracy. The missing values ​​of the data were processed using the mean interpolation method. have:

[0080]

[0081] in For missing values, use Make a substitution,

[0082] Use the standard deviation method to remove noise and use the mean and standard deviation Determine whether a data point is an outlier. If the data point The value exceeds The range is considered as noise, that is:

[0083]

[0084] Smoothing time series data by exponential weighted averaging, the simple exponential smoothing formula is:

[0085]

[0086] in is the smoothed value, is the smoothing coefficient ,

[0087] Finally, the processed data is standardized and the eigenvalues ​​are scaled to In scope:

[0088] .

[0089] (2) Label the data and mark whether each data point belongs to a fault and its fault type. The data labels of the distribution network grounding fault are shown in Table 1. The output value is the label of the fault category. The fault is set to 15 types of labels from 0 to 14, corresponding to "branch burden, tree barrier, transformer joint, fuse failure, pole switch failure, PT failure, pole underground cable failure, conductor burden, poor insulation of ground cable, ground cable discharge, cabinet cable discharge, animal intrusion in the cabinet, low voltage fault, lightning strike, and lightning conductor detachment."

[0090]

[0091] (3) By embedding the ground fault data into a high-dimensional trajectory matrix, the fault matrix is ​​subjected to singular value decomposition (SVD) to extract the main signal components.

[0092] Each fault signal data is constructed into a matrix. Assuming that the number of data points of each signal is N and the length of the signal is L, the signal data can be reconstructed into an L×K trajectory matrix X, where ,

[0093] Perform singular value decomposition on the reconstructed trajectory matrix X. The specific formula is:

[0094]

[0095] Where: is a left singular matrix, is a right singular matrix,

[0096] From the singular value matrix The first few main singular values ​​are selected, which correspond to the main features of the fault signal. By analyzing the size of these singular values, different fault types can be effectively distinguished, thereby achieving the purpose of optimizing the classification ability of LSTM.

[0097] (4) The singular value eigenvectors extracted from ISSA As input data of the LSTM model, for each time step, the LSTM model receives a set of singular value feature vectors, forming a feature sequence input:

[0098] (6)

[0099] In the formula for different time steps.

[0100] (5) Establish an LSTM model, whose structure is shown in the following figure: Figure 2 , determine the parameters of the forget gate, input gate, and output gate of the core part of the model. In the model parameter setting, the number of hidden units in the LSTM layer is set to 50, the learning rate is set to 0.001, and the cross entropy function is selected as the loss function, the formula is shown in Formula 7, and the activation function uses the tanh function, the formula is shown in Formula 8. During the model training process, iterative optimization is performed through the training set, and the validation set is used for tuning to prevent overfitting. Finally, the trained LSTM model is used to classify the grounding fault of the new data, output the probability distribution of each category, and determine the specific fault type.

[0101] (7)

[0102] is the predicted probability; is the actual label; L is the predicted result,

[0103] (8)

[0104] For any real input x, the result of tanh(x) is another real number between -1 and 1, excluding -1 and 1, which is determined by the definition of the hyperbolic tangent function, where e is the base of the natural logarithm.

[0105] The formulas for the forget gate, input gate, and output gate of the LSTM model are as follows:

[0106] The forget gate retains or deletes the information in the neuron state and filters the content that needs to be remembered.

[0107] (9)

[0108] Where: and Represents the weight between the forget gate and the input and historical output; is the current input; is the bias term; For historical output; is the output of the forget gate.

[0109] The input gate can obtain the update degree of the current information, thereby realizing the update of the unit state.

[0110] (10)

[0111] Where: and Respectively represent the weights between the input, the output of the previous moment, and the input gate; Bias term; is the current input; For historical output; is the output of the input gate.

[0112] The output gate confirms the value of the next state and determines the output state at this moment.

[0113] (11)

[0114] Where: and Represents the weight between the output gate and the input and historical output; is the bias term; is the current input; For historical output.

[0115] (6) The parameters of ISSA-LSTM are set as shown in Table 2. The newly collected unlabeled fault data is input into the pre-trained ISSA-LSTM model, the model is trained repeatedly, and the accuracy of the model in fault classification is verified to form the final ISSA-LSTM model.

[0116] Table 2 ISSA-LSTM parameter setting table

[0117]

[0118] (7) Use the test set data to verify the ISSA-LSTM model and evaluate the performance of the model to ensure the effectiveness and reliability of the model in a real environment.

[0119] By introducing the improved sparrow search algorithm (ISSA) to optimize the long short-term memory network (LSTM), the shortcomings of the distribution network grounding fault classification are improved. Its goals include improving the accuracy of fault classification, improving the model's ability to identify complex fault modes by optimizing hyperparameters; enhancing real-time monitoring capabilities to achieve rapid response and dynamic adjustment; handling diverse fault types, integrating multi-dimensional data to improve the comprehensiveness of classification; reducing false alarm rates, reducing false alarms and missed alarms by optimizing the learning process; and improving training efficiency, accelerating model convergence, and providing a quick solution for practical applications. The results are shown in the following table.

[0120] Table 3 Diagnostic accuracy of different optimization algorithms

[0121]

[0122]

[0123] The embodiment of the present application provides a distribution network grounding fault classification system, which includes: a memory and a processor, wherein the memory includes a program of a distribution network grounding fault classification method, and when the program of the distribution network grounding fault classification method is executed by the processor, the following steps are implemented: collecting a variety of distribution network grounding fault raw data and preprocessing the data; label encoding the data, marking whether each data point belongs to a fault and its fault type; embedding the grounding fault data into a high-dimensional trajectory matrix through an improved sparrow search algorithm, performing singular value decomposition on the fault matrix, and extracting the main signal components; using the singular value feature vector extracted from ISSA as the input data of the LSTM model; establishing the LSTM model, and determining the parameters of the forgetting gate, input gate, and output gate of the core part of the model; setting the corresponding parameters of the ISSA algorithm and the initial hyperparameters of the LSTM; using ISSA to optimize the hyperparameters in the LSTM network, and combining them to form an ISSA-LSTM model; and using the data of the test set to verify the ISSA-LSTM model.

[0124] An embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores program codes, and when the program codes are executed by a processor, the steps of the distribution network grounding fault classification method as described above are implemented.

[0125] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0126] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0127] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0128] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0129] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0130] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0131] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0132] The above description is only an embodiment of the present application and is not intended to limit the protection scope of the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A distribution network grounding fault classification method, characterized in that: The following steps are involved: Collect various raw data of distribution network grounding faults and pre-process the data; Label the data and mark whether each data point is a fault and its fault type; The ground fault data is embedded into a high-dimensional trajectory matrix through an improved sparrow search algorithm, and the fault matrix is ​​subjected to singular value decomposition to extract the main signal components. The singular value feature vector extracted from ISSA is used as the input data of the LSTM model; Establish an LSTM model and determine the parameters of the forget gate, input gate, and output gate in the core part of the model; Set the corresponding parameters of the ISSA algorithm and the initial hyperparameters of the LSTM; Use ISSA to optimize the hyperparameters in the LSTM network and combine them to form an ISSA-LSTM model; The ISSA-LSTM model is validated using the data from the test set.

2. A distribution network grounding fault classification method according to claim 1, characterized in that: The method of collecting a variety of distribution network grounding fault raw data and preprocessing the data is as follows: The original data of 15 types of ground faults in distribution networks were collected and preprocessed, including processing missing values, removing noise, smoothing data, and standardizing data to ensure data consistency and accuracy. The missing values ​​of the data were processed using the mean interpolation method. have: （1） in For missing values, use Make a substitution, Use the standard deviation method to remove noise and use the mean and standard deviation Determine whether a data point is an outlier. If the data point The value exceeds The range is considered as noise, that is: （2） Smooth the time series data by exponential weighted average. The simple exponential smoothing formula is: （3） in is the smoothed value, is the smoothing coefficient , Finally, the processed data is standardized and the eigenvalues ​​are scaled to In scope: （4）。 3. A distribution network grounding fault classification method according to claim 1, characterized in that: The data is labeled and encoded to mark whether each data point belongs to a fault and its fault type. Specifically, the fault is set to 15 types of labels from 0 to 14, corresponding to tree branches, tree obstacles, transformer joints, fuse failure, pole switch failure, PT failure, pole underground cable failure, wire burden, poor insulation of ground cable, ground cable discharge, cabinet cable discharge, animal intrusion into the cabinet, low voltage fault, lightning strike, and lightning conductor detachment.

4. A distribution network grounding fault classification method according to claim 1, characterized in that: The improved sparrow search algorithm is used to embed the ground fault data into a high-dimensional trajectory matrix, perform singular value decomposition on the fault matrix, and extract the main signal components, specifically: Each fault signal data is constructed into a matrix. Assuming that the number of data points of each signal is N and the length of the signal is L, the signal data can be reconstructed into an L×K trajectory matrix X, where , Perform singular value decomposition on the reconstructed trajectory matrix X. The specific formula is: （5） Where: is a left singular matrix, is a right singular matrix, From the singular value matrix The first few main singular values ​​are selected, which correspond to the main features of the fault signal. By analyzing the size of these singular values, different fault types can be effectively distinguished, thereby achieving the purpose of optimizing the classification ability of LSTM.

5. A distribution network grounding fault classification method according to claim 1, characterized in that: The singular value feature vector extracted from ISSA is used as the input data of the LSTM model as follows: The singular value eigenvectors extracted from the ISSA As input data of the LSTM model, for each time step, the LSTM model receives a set of singular value feature vectors, forming a feature sequence input: （6） In the formula for different time steps.

6. A distribution network grounding fault classification method according to claim 1, characterized in that: The LSTM model is established, and the parameters of the forget gate, input gate, and output gate of the core part of the model are determined as follows: in the model parameter setting, the number of hidden units of the LSTM layer is set to 50, the learning rate is set to 0.001, and the cross entropy function is selected as the loss function, and the activation function adopts the tanh function. During the model training process, iterative optimization is performed through the training set, and the validation set is used for tuning to prevent overfitting. Finally, the trained LSTM model is used to classify the grounding faults of the new data, and the probability distribution of each category is output to determine the specific fault type.

7. A distribution network grounding fault classification method according to claim 1, characterized in that: The parameters corresponding to the ISSA algorithm and the LSTM initial hyperparameters are specifically set as follows: The parameters of the ISSA-LSTM model are set, the newly collected unlabeled fault data is input into the pre-trained ISSA-LSTM model, the model is trained repeatedly, and the accuracy of the model in fault classification is verified to form the final ISSA-LSTM model.

8. A distribution network grounding fault classification method according to claim 1, characterized in that: The method of using ISSA to optimize the hyperparameters in the LSTM network and combining them to form the ISSA-LSTM model specifically includes: using ISSA to optimize the hyperparameters of the LSTM, firstly, generating an initial population, each individual representing a set of LSTM hyperparameters, then training the LSTM model and evaluating the performance of each set of parameters, and updating the population according to the fitness evaluation results, generating new individuals, and selecting individuals with higher fitness for the next round of iterations, and finally, iteratively executing the above steps until the maximum number of iterations is reached or the stopping condition is met.

9. A distribution network ground fault classification system, characterized in that: The system comprises: a memory and a processor, wherein the memory comprises a program of a distribution network ground fault classification method, and when the program of the distribution network ground fault classification method is executed by the processor, the following steps are implemented: collecting a variety of distribution network ground fault raw data and preprocessing the data; label encoding the data, marking whether each data point belongs to a fault and its fault type; embedding the ground fault data into a high-dimensional trajectory matrix through an improved sparrow search algorithm, performing singular value decomposition on the fault matrix, and extracting the main signal components; using the singular value feature vector extracted from ISSA as the input data of the LSTM model; establishing the LSTM model, determining the parameters of the forgetting gate, input gate, and output gate of the core part of the model; setting the corresponding parameters of the ISSA algorithm and the initial hyperparameters of the LSTM; using ISSA to optimize the hyperparameters in the LSTM network, and combining to form an ISSA-LSTM model; and using the data of the test set to verify the ISSA-LSTM model.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores program codes, and when the program codes are executed by a processor, the steps of the method for classifying ground faults in a distribution network as claimed in any one of claims 1 to 8 are implemented.

Images