Decision basis analysis method, apparatus and device for high-impedance grounding fault identification model

Abstract

A decision basis analysis method, apparatus and device for a high-impedance grounding fault identification model. The method comprises: on the basis of unlabeled zero-sequence current data, training an initial fault identification model to obtain a preset high-impedance grounding fault identification model, wherein the preset high-impedance grounding fault identification model comprises an encoder and a decoder; analyzing the degree of importance of each encoded feature, which is output by the encoder, to an identification result of the model, and calculating a corresponding global Shapley value; using an instance normalization algorithm to perform normalization processing on encoded vectors, and then inputting the normalized encoded vectors into the decoder for decoding analysis, so as to obtain a decoded waveform; and performing spectrum analysis by means of comparing the decoded waveform with an original fault waveform, and analyzing a decision basis of the model on the basis of the global Shapley values, so as to obtain an identification basis of the model. The present application can solve the technical problems of existing grounding fault identification models requiring a large amount of labelled data and the models lacking interpretability, thus leading to a lack of specificity and reliability in model training.

Description

Decision-making basis analysis method, device and equipment for high-resistance grounding fault identification model

[0001] This application claims priority to the Chinese patent application filed with the China Patent Office on March 22, 2024, with application number 202410341908.2 and invention name “Decision-making basis analysis method, device and equipment for high-resistance grounding fault identification model”, the entire contents of which are incorporated by reference into this application. Technical Field

[0002] The present application relates to the technical field of distribution network fault analysis, and in particular to a decision basis analysis method, device and equipment for a high-resistance grounding fault identification model. Background Art

[0003] As distribution network structures become increasingly complex, the probability of failure increases. Single-phase ground faults are one of the most common fault types in distribution networks. When a single-phase ground fault occurs in a distribution network, it is often accompanied by arcing and generates an arc-to-ground overvoltage. This voltage has a high amplitude. If the voltage persists for a long time, and the capacitive current increases with the increase in feeders, the overvoltage generated during high-resistance ground fault operation can easily cause new grounding points in system equipment, leading to phase-to-phase short circuits or two-point or multiple-point ground faults, further escalating the incident. Therefore, accurate fault identification is crucial to the stable operation of distribution networks.

[0004] The main methods for identifying high-resistance grounding faults in power grids include time domain, frequency domain, time-frequency domain, and artificial intelligence. The time domain method focuses on the unique time-domain characteristics of voltage and current signals, which have distinct physical properties. The frequency domain method uses the high- and low-frequency components of voltage and current signals to distinguish high-resistance grounding faults from external interference. The intelligent identification method for high-resistance grounding faults (HIFs), which combines signal processing technology with artificial intelligence algorithms, effectively processes massive amounts of data through adaptive learning of deep features, avoiding the limitations of manual fault feature extraction due to prior experience.

[0005] However, the AI-based HIF identification method not only requires a large amount of labeled data, but also the training process is in a black box mode, resulting in a lack of interpretability of the model. Therefore, it is difficult to make a scientific and reliable analysis of high-resistance grounding fault identification, and it cannot provide effective decision-making guidance for the training of fault identification models in specific scenarios.

[0006] Summary of the Invention

[0007] This application provides a decision basis analysis method, device and equipment for a high-resistance grounding fault identification model, which is used to solve the technical problems that the existing grounding fault identification model has a large demand for labeled data and the model lacks interpretability, resulting in a lack of pertinence and reliability in model training.

[0008] In view of this, the first aspect of the present application provides a decision basis analysis method for a high-resistance grounding fault identification model, including:

[0009] Training an initial fault recognition model based on unlabeled zero-sequence current data to obtain a preset high-resistance grounding fault recognition model, wherein the preset high-resistance grounding fault recognition model includes an encoder and a decoder;

[0010] Analyzing the importance of the encoding features output by the encoder to the recognition result of the preset high-resistance grounding fault recognition model, and calculating the corresponding global Shapley value;

[0011] After normalizing the coding vector composed of the plurality of coding features using an instance normalization algorithm, the coding vector is input into the decoder for decoding analysis to obtain a decoding waveform;

[0012] The spectrum analysis is performed by comparing the decoded waveform with the original fault waveform, and the model decision basis is analyzed based on the global Shapley value to obtain the model identification basis.

[0013] Preferably, the training of the initial fault identification model based on the unlabeled zero-sequence current data to obtain a preset high-resistance grounding fault identification model includes:

[0014] Obtain zero-sequence current data in the distribution network to obtain label-free zero-sequence current data;

[0015] Pre-training an initial encoder and an initial decoder using the unlabeled zero-sequence current data to obtain an encoder and a decoder;

[0016] A nonlinear mapping is established between the encoder, the decoder and a high-resistance grounding fault label of a preset fully connected network layer to generate a preset high-resistance grounding fault identification model.

[0017] Preferably, the method of pre-training an initial encoder and an initial decoder using the unlabeled zero-sequence current data to obtain an encoder and a decoder further includes:

[0018] Calculate the fitting error between the input zero-sequence current wave and the decoded reconstructed wave;

[0019] Calculate the waveform similarity between the input zero-sequence current wave and the decoded reconstructed wave;

[0020] An objective function of the encoder is constructed according to the fitting error and the waveform similarity.

[0021] Preferably, analyzing the importance of the coding feature output by the encoder to the recognition result of the preset high-resistance grounding fault recognition model and calculating the corresponding global Shapley value includes:

[0022] Constructing an additive explanatory model based on the Shapley value of game theory;

[0023] Analyzing the importance of each encoding feature of each sample output by the encoder to the model recognition result according to the additive interpretation model to obtain a local Shapley value;

[0024] The absolute average of the local Shapley values ​​corresponding to the coding features at the same position of all samples is calculated to obtain a global Shapley value.

[0025] Preferably, the analysis of the importance of the coding features output by the encoder to the model recognition result and calculation of the corresponding global Shapley value, after the plurality of coding features form a coding vector, further includes:

[0026] Arrange the global Shapley values ​​of all samples in descending order to obtain a Shapley value sequence.

[0027] The second aspect of the present application provides a decision basis analysis device for a high-resistance grounding fault identification model, comprising:

[0028] A model generation unit is used to train an initial fault recognition model based on unlabeled zero-sequence current data to obtain a preset high-resistance grounding fault recognition model, wherein the preset high-resistance grounding fault recognition model includes an encoder and a decoder;

[0029] a feature analysis unit, configured to analyze the importance of the encoding feature output by the encoder to the recognition result of the preset high-resistance grounding fault recognition model, and calculate the corresponding global Shapley value;

[0030] A decoding processing unit, configured to normalize the coding vector composed of the plurality of coding features using an instance normalization algorithm, and then input the normalized vector into the decoder for decoding analysis to obtain a decoded waveform;

[0031] The basis analysis unit is used to perform spectrum analysis by comparing the decoded waveform and the original fault waveform, and analyze the model decision basis based on the global Shapley value to obtain the model identification basis.

[0032] Preferably, the model generating unit is specifically used to:

[0033] Obtain zero-sequence current data in the distribution network to obtain label-free zero-sequence current data;

[0034] Pre-training an initial encoder and an initial decoder using the unlabeled zero-sequence current data to obtain an encoder and a decoder;

[0035] A nonlinear mapping is established between the encoder, the decoder and a high-resistance grounding fault label of a preset fully connected network layer to generate a preset high-resistance grounding fault identification model.

[0036] Preferably, it further comprises: a training function construction unit, specifically used for:

[0037] Calculate the fitting error between the input zero-sequence current wave and the decoded reconstructed wave;

[0038] Calculate the waveform similarity between the input zero-sequence current wave and the decoded reconstructed wave;

[0039] An objective function of the encoder is constructed according to the fitting error and the waveform similarity.

[0040] Preferably, the feature analysis unit is specifically used to:

[0041] Constructing an additive explanatory model based on the Shapley value of game theory;

[0042] Analyzing the importance of each encoding feature of each sample output by the encoder to the model recognition result according to the additive interpretation model to obtain a local Shapley value;

[0043] The absolute average of the local Shapley values ​​corresponding to the coding features at the same position of all samples is calculated to obtain a global Shapley value.

[0044] A third aspect of the present application provides a decision basis analysis device for a high-resistance grounding fault identification model, the device comprising a processor and a memory;

[0045] The memory is used to store program code and transmit the program code to the processor;

[0046] The processor is configured to execute the decision basis analysis method of the high-resistance grounding fault identification model described in the first aspect according to the instructions in the program code.

[0047] It can be seen from the above technical solutions that the embodiments of the present application have the following advantages:

[0048] The present application provides a decision basis analysis method for a high-resistance grounding fault identification model, including: training an initial fault identification model based on unlabeled zero-sequence current data to obtain a preset high-resistance grounding fault identification model, wherein the preset high-resistance grounding fault identification model includes an encoder and a decoder; analyzing the importance of the coding features output by the encoder to the identification result of the preset high-resistance grounding fault identification model, and calculating the corresponding global Shapley value; normalizing a coding vector composed of multiple coding features using an instance normalization algorithm, and inputting the normalized data into a decoder for decoding analysis to obtain a decoded waveform; performing spectral analysis by comparing the decoded waveform with the original fault waveform, and analyzing the model decision basis based on the global Shapley value to obtain a model identification basis.

[0049] The decision-making basis analysis method for the high-resistance ground fault identification model provided in this application uses unlabeled zero-sequence current data to perform unsupervised pre-training on the encoder model, and based on this, constructs a model for high-resistance ground fault identification. The model does not require a high amount of labeled input data, but can still meet the needs of fault identification. In addition, the model is interpreted based on the Shapley value and instance normalization algorithm to determine the basis for the model's fault identification analysis. This can provide a targeted and reliable reference theoretical support for model training, thereby improving the accuracy and credibility of model identification. Therefore, this application can solve the technical problems that existing ground fault identification models have a large demand for labeled data and the model lacks interpretability, resulting in a lack of targetedness and reliability in model training. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] FIG1 is a flow chart of a decision basis analysis method for a high-resistance ground fault identification model provided by an embodiment of the present application;

[0051] FIG2 is a schematic structural diagram of a decision basis analysis device for a high-resistance ground fault identification model provided by an embodiment of the present application;

[0052] FIG3 is a diagram illustrating an example structure of a preset high-resistance ground fault identification model provided in an embodiment of the present application;

[0053] FIG4 is a schematic diagram of the circuit structure of a radial distribution network model provided in an application example of this application;

[0054] FIG5 is a schematic diagram of an encoder network structure provided by an application example of this application;

[0055] FIG6 is a global Shapley value ranking diagram corresponding to different feature vectors provided in the application example of this application;

[0056] FIG7 is a schematic diagram of the overall framework of the explainable high-resistance fault identification model provided in the application example of this application. DETAILED DESCRIPTION

[0057] In order to help those skilled in the art better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0058] For ease of understanding, please refer to FIG1 , which shows an embodiment of a decision basis analysis method for a high-resistance ground fault identification model provided by this application, including:

[0059] Step 101: Train an initial fault recognition model based on unlabeled zero-sequence current data to obtain a preset high-resistance grounding fault recognition model, where the preset high-resistance grounding fault recognition model includes an encoder and a decoder.

[0060] Furthermore, step 101 includes:

[0061] Obtain zero-sequence current data in the distribution network to obtain label-free zero-sequence current data;

[0062] The initial encoder and decoder are pre-trained using unlabeled zero-sequence current data to obtain an encoder and a decoder;

[0063] A nonlinear mapping between the encoder, decoder and the high-resistance grounding fault label of the preset fully connected network layer is established to generate a preset high-resistance grounding fault identification model.

[0064] Furthermore, the initial encoder and the initial decoder are pre-trained using unlabeled zero-sequence current data to obtain an encoder and a decoder, further comprising:

[0065] Calculate the fitting error between the input zero-sequence current wave and the decoded reconstructed wave;

[0066] Calculate the waveform similarity between the input zero-sequence current wave and the decoded reconstructed wave;

[0067] The objective function of the encoder is constructed based on the fitting error and waveform similarity.

[0068] It should be noted that the preset high-resistance grounding fault identification model of this embodiment is mainly composed of an encoder, a decoder and a preset fully connected network layer, and adopts a pre-training method of model migration for unsupervised training. During this process, a large amount of zero-sequence current data in the distribution network is relatively easy to obtain, so that the encoder can learn unsupervisedly and capture the key features of the zero-sequence current data of the distribution network, and generate a compact encoding; in this way, the model can have a universal feature extraction capability for relevant data of the distribution network; then, a nonlinear mapping between the encoding features and the high-resistance grounding fault label is established through the preset fully connected network layer to generate the preset high-resistance grounding fault identification model.

[0069] The model's encoder converts the input data into an encoding in the latent representation space, while the decoder attempts to restore this encoding to the original input. Through this process, the autoencoder reconstructs the input data, aiming to minimize the reconstruction error—that is, to ensure that the decoded data is as close to the original input as possible. By training the autoencoder with the goal of minimizing the difference between the input data waveform and its reconstructed waveform, we can construct the encoder's training objective function:

[0070] F=SSE+SW

[0071] Among them, F is the difference between the overall input data and the reconstructed waveform, SSE is the fitting error, and SW is the waveform similarity. The calculation process of the two is:

[0072] Among them, i f 、i s are the input waveform data and the reconstructed waveform data, k, N T / 2 are the kth sampling point and the total number of sampling times within half the power frequency cycle, ω k is the calculation weight of the kth sampling point. The value range of SW is [0,1]. The value indicates the degree of similarity. The smaller the value, the higher the similarity between the two waveforms.

[0073] After using an autoencoder to generate a compact code that captures the key features of the zero-sequence current data, the model then establishes a nonlinear mapping from the code features to high-resistance ground fault labels. This is accomplished in a pre-set fully connected network layer. Furthermore, supervised training of the fully connected network layer using a small amount of labeled data enables a complex mapping between the code and the high-resistance ground fault labels, resulting in a highly accurate pre-set high-resistance ground fault identification model.

[0074] Because the construction and use of the preset high-resistance ground fault identification model are both black-box processes, users cannot determine the specific data or characteristics of the distribution network that the model uses to analyze the fault and obtain the identification results. This naturally prevents targeted model adjustments and flexible application based on actual scenario characteristics. Therefore, this embodiment provides an analytical solution to interpret the model, identify the decision basis for the model's high-resistance ground fault identification, and better guide the model's training and application in specific scenarios.

[0075] Step 102: Analyze the importance of the encoding features output by the encoder to the recognition result of the preset high-resistance grounding fault recognition model, and calculate the corresponding global Shapley value.

[0076] Furthermore, step 102 includes:

[0077] Constructing an additive explanatory model based on the Shapley value of game theory;

[0078] The importance of each encoding feature of each sample output by the encoder to the model recognition result is analyzed based on the additive interpretation model to obtain the local Shapley value;

[0079] The absolute average of the local Shapley values ​​corresponding to the coding features at the same position of all samples is calculated to obtain the global Shapley value.

[0080] Furthermore, step 102 further includes:

[0081] Arrange the global Shapley values ​​of all samples in descending order to obtain a Shapley value sequence.

[0082] It should be noted that the Shapley value is originally used in game theory to evaluate the contribution to the benefits. This embodiment calculates the Shapley value of the encoded feature vector extracted from each sample to analyze its importance to the model prediction result.

[0083] The core idea is to sample the additive explanatory model g(x) constructed based on the Shapley value of game theory to fit the complex model f(x), that is, the preset high-resistance grounding fault identification model of this embodiment, and provide a feasible explanation scheme for the prediction results of the model. The specific expression is:

[0084] Among them, n is the number of features of sample x, φ0 is the model's prediction benchmark value for the sample, and is also the mean of the model's prediction results for all samples, φ j is the jth feature x of sample x j The Shapley value of quantified feature x j The degree of influence on the model recognition results.

[0085] Therefore, the model's prediction result for any sample can be expressed as the sum of the prediction baseline value and the Shapley values ​​of all the coding features of the sample. The coding feature vector of the sample can correspond to multiple influence values, which can be simply understood as multiple Shapley values. The Shapley value is used to measure the importance of the coding feature to the model's prediction results, reflecting the contribution of the coding feature to the model's prediction results. Its specific calculation principle can be expressed as:

[0086] Among them, {x 1 ,x 2 ,......,x n} represents the coded feature vector of n-dimensional sample x, that is, the feature set, which contains n coded features, and S does not contain the coded feature x j The characteristic subset of |S| is the number of characteristic elements of set S, ∪ is the union operation, f x (S∪{x j}),f x (S) respectively represent the encoding features x j and does not contain the encoded feature x j The prediction results of the model under the condition of φ j The first part on the right side of the calculation formula represents the weight. Encoded feature x j The Shapley value φ j Defined as the mean of its marginal contribution in different subsets of encoded features.

[0087] For any given sample, the Shapley value of each coded feature reflects the degree of influence of the feature on the prediction result and the positive or negative influence. That is, the feature with a larger absolute value of the Shapley value has a greater impact on the model prediction result. The positive or negative value of the Shapley value indicates that the feature will increase or decrease the output result of the model, which greatly increases the local interpretability of the complex model f(x). That is, the Shapley value calculated for a single coded feature belongs to the local Shapley value. Based on this, if a feature x of all samples j The absolute value of the Shapley value is taken as the average value to measure the global importance of the feature, which can be used to globally explain the complex model f(x). Therefore, the calculation process of the global Shapley value is expressed as:

[0088] Among them, φ j (x i ) is the i-th sample x i The j-th encoded feature x j The Shapley value, that is, the local Shapley value, Y jis the jth encoding feature x j The corresponding global Shapley value, m is the total number of samples.

[0089] Evaluating the influence of coding features on the model fault identification results based on Shapley values ​​can greatly increase the interpretability and credibility of the model. This embodiment can also sort the global Shapley values ​​in descending order to analyze the coding features that have the greatest impact on the model prediction results.

[0090] Step 103: After normalizing the coding vector composed of multiple coding features using an instance normalization algorithm, the code is input into a decoder for decoding analysis to obtain a decoded waveform.

[0091] Using the instance normalization algorithm to process each vector composed of coding features one by one, the mean of the corresponding coding feature vector can be reduced to 0 and the variance can be normalized to 1. Please refer to Figure 3. After the instance normalization process, the features expressed by the vector in the code can be removed, because the dimension of each vector represents a coding feature extracted by a 1-DCNN. Assuming that the rectangular box in Figure 3 is the third harmonic feature, the third harmonic feature of HIF will be obvious, and the third harmonic feature of Non-HIF will not be obvious. If these features are removed after instance normalization, they can be removed. Instance normalization performs independent normalization on each sample of the coding feature vector. The specific instance normalization process can be expressed as:

[0092] Where c is the number of vectors, N is the number of eigenvectors, μ c is the mean of the current eigenvector, σ c is the standard deviation of the current eigenvector, x ic is the current encoding feature vector, is the new encoded feature vector after instance normalization, and ε is a small constant, usually 1e -5 , or a smaller value. The new encoded feature vector after instance normalization is input into the decoder for decoding, and the output waveform of the decoder, that is, the decoded waveform, can be obtained.

[0093] Step 104 : Perform spectrum analysis by comparing the decoded waveform with the original fault waveform, and analyze the model decision basis based on the global Shapley value to obtain the model identification basis.

[0094] The original fault waveform is the waveform data input to the encoder, which is used to reflect the data of the high-resistance grounding fault. The decoded waveform reconstructed by the decoder is compared with the original fault waveform to analyze the spectral differences. Attribution analysis is performed based on the global Shapley value to determine the contribution of the decoding vector that does not pass through the input preset fully connected network layer to the prediction and recognition results. Combined with the fault characteristics in the frequency domain when the high-resistance grounding fault (HIF) test occurs, the recognition decision basis of the pre-trained high-resistance grounding fault recognition model based on the feature extractor of the autoencoder and the fully connected neural network classifier is explained.

[0095] For ease of understanding, this application also provides an application example of the decision-making basis analysis method of the high-resistance grounding fault identification model. Please refer to Figure 4. The radial distribution network model is established using EMTDC / PSCAD simulation software. The system power frequency is 50 Hz, the sampling rate is 4 kHz, and the parameters of the cable line and overhead line are shown in Table 1.

[0096] Table 1 Line parameters

[0097] Table 2 describes HIF or HIF interference events occurring at different fault locations (FP) and along different lines. The data in Table 2 was divided into unlabeled and labeled training sets at a ratio of 9:1. 1,800 unlabeled data samples were used to pre-train the autoencoder, and 200 labeled data samples were used to train and adjust the fully connected neural network. Capacitor switching (CS) is simulated using a parallel three-phase capacitor model, magnetizing inrush current (IC) is simulated using an unloaded single-phase transformer, low-impedance faults (LIF) are simulated using a low-resistance model (5Ω-100Ω), load switching (LS) uses a three-phase asymmetric load model, and the HIF model is the Emanuel model. Furthermore, to align with engineering practice, asynchronous CS closing was added to the experiment to simulate non-fault transients. In this study, three-phase asynchronous closing means that phase A is connected to the system first, followed by phases B and C simultaneously with the same delay. The initial fault angle was set to 0°, 30°, 60°, 90°, and 120°.

[0098] Table 2 HIF and disturbance event samples

[0099] The structure of the pretrained high-resistance ground fault identification model for this application example is shown in Figure 3. The fully connected neural network has a structure of 108 × 138 × 168 × 38 × 2. Specifically, the encoder is a 1-DCNN, as shown in Figure 5. The decoder has a symmetrical structure, and its parameters are shown in Table 3.

[0100] Table 3 Autoencoder related parameters

[0101] This application example uses a 1-DCNN to directly operate on the raw zero-sequence current signal. This effectively captures local features within the signal and, through convolution operations, learns patterns and variations between adjacent time steps in the input signal. This enables the model to better understand temporal relationships within the signal. The 1-DCNN uses sliding convolution kernels, reducing the number of parameters required for learning through parameter sharing. This means the model can learn patterns and structures within the signal with a smaller parameter size, improving training efficiency and generalization performance. After training the autoencoder, the encoder portion is extracted to serve as the feature extraction module of the recognition model. Its input is a 200×1 zero-sequence current waveform, and its output is an 18×6 feature vector. The fully connected neural network's input layer has 108 neurons, connecting the autoencoder's encoding output to the hidden layer. Experiments have shown that a hidden layer with a 138×168×38 configuration achieves good recognition results for measured high-resistance data. The output layer of the neural network has two neurons, one corresponding to high-resistance and the other to non-high-resistance.

[0102] The Shapley value is used to analyze the influence of a vector on the model's prediction results. If the input vector of the fully connected neural network is 6-dimensional, please refer to Figure 6. The horizontal axis in the figure represents the global Shapley value of the feature, that is, the average influence of the input vector on the model's prediction results. Different colors represent the average influence of the feature on different model prediction results. The vertical axis is the feature name. In short, the larger the global Shapley value of a feature, the greater the contribution of the corresponding input vector to the model's prediction results and the higher its importance. Therefore, input vectors D4, D5, and D6 contribute more to the model's prediction results and are more important. Use instance normalization to process each input vector of the fully connected neural network one by one, reducing the mean of the corresponding input vector to 0 and standardizing the variance to 1, removing the features expressed by the vector in the encoding.

[0103] The instance-normalized encoded vector is input into the decoder of the original autoencoder to obtain a new output waveform. Spectral analysis is then performed on the original and instance-normalized waveforms to compare their frequency domain differences. See Figure 7 for the overall process. Global Shapley value attribution analysis reveals that the D4 vector contributes most significantly to the fully connected neural network prediction results. The third harmonic component exhibits the largest spectral difference after erasing D4. This, combined with electrical engineering expertise, explains that one of the key decision factors for detecting HIF in the recognition model based on the autoencoder's feature extractor and the fully connected neural network classifier is the detection of differences in the third harmonic component.

[0104] The decision-making basis analysis method for the high-resistance ground fault identification model provided in the embodiments of the present application uses unlabeled zero-sequence current data to perform unsupervised pre-training on the encoder model, and based on this, constructs a model for high-resistance ground fault identification. The model does not require a high amount of labeled input data, but can still meet the needs of fault identification. In addition, the model is interpreted based on Shapley values ​​and instance normalization algorithms to determine the basis for the model's fault identification analysis. This provides a targeted and reliable reference theoretical support for model training, thereby improving the accuracy and credibility of model identification. Therefore, the embodiments of the present application can solve the technical problems of existing ground fault identification models, such as the large amount of labeled data required and the lack of interpretability of the model, which leads to a lack of targetedness and reliability in model training.

[0105] For ease of understanding, please refer to FIG2 . This application provides an embodiment of a decision basis analysis device for a high-resistance grounding fault identification model, including:

[0106] A model generation unit 201 is configured to train an initial fault recognition model based on unlabeled zero-sequence current data to obtain a preset high-resistance grounding fault recognition model, wherein the preset high-resistance grounding fault recognition model includes an encoder and a decoder;

[0107] The feature analysis unit 202 is used to analyze the importance of the encoding feature output by the encoder to the recognition result of the preset high-resistance grounding fault recognition model, and calculate the corresponding global Shapley value;

[0108] The decoding processing unit 203 is configured to normalize the code vector composed of multiple code features using an example normalization algorithm, and then input the normalized code vector into a decoder for decoding analysis to obtain a decoded waveform.

[0109] The basis analysis unit 204 is used to perform spectrum analysis by comparing the decoded waveform with the original fault waveform, and analyze the model decision basis based on the global Shapley value to obtain the model identification basis.

[0110] Furthermore, the model generation unit 201 is specifically configured to:

[0111] Obtain zero-sequence current data in the distribution network to obtain label-free zero-sequence current data;

[0112] The initial encoder and decoder are pre-trained using unlabeled zero-sequence current data to obtain an encoder and a decoder;

[0113] A nonlinear mapping between the encoder, decoder and the high-resistance grounding fault label of the preset fully connected network layer is established to generate a preset high-resistance grounding fault identification model.

[0114] Furthermore, the system further includes a training function construction unit 205, which is specifically configured to:

[0115] Calculate the fitting error between the input zero-sequence current wave and the decoded reconstructed wave;

[0116] Calculate the waveform similarity between the input zero-sequence current wave and the decoded reconstructed wave;

[0117] The objective function of the encoder is constructed based on the fitting error and waveform similarity.

[0118] Furthermore, the feature analysis unit 202 is specifically configured to:

[0119] Constructing an additive explanatory model based on the Shapley value of game theory;

[0120] The importance of each encoding feature of each sample output by the encoder to the model recognition result is analyzed based on the additive interpretation model to obtain the local Shapley value;

[0121] The absolute average of the local Shapley values ​​corresponding to the coding features at the same position of all samples is calculated to obtain the global Shapley value.

[0122] The present application also provides a decision basis analysis device for a high-resistance grounding fault identification model, characterized in that the device includes a processor and a memory;

[0123] The memory is used to store program codes and transmit the program codes to the processor;

[0124] The processor is configured to execute the decision basis analysis method of the high-resistance grounding fault identification model in the above method embodiment according to the instructions in the program code.

[0125] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0126] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0127] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0128] If the integrated unit 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 application is essentially or the part that contributes to the prior art or all or 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 and includes a number of instructions for executing all or part of the steps of the method described in each embodiment of the present application through a computer device (which can be a personal computer, server, or network device, etc.). The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (English full name: Read-Only Memory, English abbreviation: ROM), random access memory (English full name: Random Access Memory, English abbreviation: RAM), disk or optical disk and other media that can store program code.

[0129] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. The decision basis analysis method of the high-resistance grounding fault identification model is characterized by: include: Training an initial fault recognition model based on unlabeled zero-sequence current data to obtain a preset high-resistance grounding fault recognition model, wherein the preset high-resistance grounding fault recognition model includes an encoder and a decoder; Analyzing the importance of the encoding features output by the encoder to the recognition result of the preset high-resistance grounding fault recognition model, and calculating the corresponding global Shapley value; After normalizing the coding vector composed of the plurality of coding features using an instance normalization algorithm, the coding vector is input into the decoder for decoding analysis to obtain a decoding waveform; The spectrum analysis is performed by comparing the decoded waveform with the original fault waveform, and the model decision basis is analyzed based on the global Shapley value to obtain the model identification basis.

2. The decision basis analysis method for the high-resistance grounding fault identification model according to claim 1 is characterized in that: The initial fault identification model is trained based on the unlabeled zero-sequence current data to obtain a preset high-resistance grounding fault identification model, including: Obtain zero-sequence current data in the distribution network to obtain label-free zero-sequence current data; Pre-training an initial encoder and an initial decoder using the unlabeled zero-sequence current data to obtain an encoder and a decoder; A nonlinear mapping is established between the encoder, the decoder and a high-resistance grounding fault label of a preset fully connected network layer to generate a preset high-resistance grounding fault identification model.

3. The decision basis analysis method for the high-resistance grounding fault identification model according to claim 2 is characterized in that: The method further comprises: pre-training an initial encoder and an initial decoder using the unlabeled zero-sequence current data to obtain an encoder and a decoder; Calculate the fitting error between the input zero-sequence current wave and the decoded reconstructed wave; Calculate the waveform similarity between the input zero-sequence current wave and the decoded reconstructed wave; An objective function of the encoder is constructed according to the fitting error and the waveform similarity.

4. The decision basis analysis method for the high-resistance grounding fault identification model according to claim 1 is characterized in that: The analyzing the importance of the encoding feature output by the encoder to the recognition result of the preset high-resistance grounding fault recognition model and calculating the corresponding global Shapley value includes: Constructing an additive explanatory model based on the Shapley value of game theory; Analyzing the importance of each encoding feature of each sample output by the encoder to the model recognition result according to the additive interpretation model to obtain a local Shapley value; The absolute average of the local Shapley values ​​corresponding to the coding features at the same position of all samples is calculated to obtain a global Shapley value.

5. The decision basis analysis method for the high-resistance grounding fault identification model according to claim 4 is characterized in that: The analysis of the importance of the coding features output by the encoder to the model recognition result and calculation of the corresponding global Shapley value, after the plurality of coding features form a coding vector, further includes: Arrange the global Shapley values ​​of all samples in descending order to obtain a Shapley value sequence.

6. The decision basis analysis device of the high resistance ground fault identification model is characterized by: include: A model generation unit is used to train an initial fault recognition model based on unlabeled zero-sequence current data to obtain a preset high-resistance grounding fault recognition model, wherein the preset high-resistance grounding fault recognition model includes an encoder and a decoder; a feature analysis unit, configured to analyze the importance of the encoding feature output by the encoder to the recognition result of the preset high-resistance grounding fault recognition model, and calculate the corresponding global Shapley value; A decoding processing unit, configured to normalize the coding vector composed of the plurality of coding features using an instance normalization algorithm, and then input the normalized vector into the decoder for decoding analysis to obtain a decoded waveform; The basis analysis unit is used to perform spectrum analysis by comparing the decoded waveform and the original fault waveform, and analyze the model decision basis based on the global Shapley value to obtain the model identification basis.

7. The decision basis analysis device for the high-resistance grounding fault identification model according to claim 6, characterized in that: The model generation unit is specifically used to: Obtain zero-sequence current data in the distribution network to obtain label-free zero-sequence current data; Pre-training an initial encoder and an initial decoder using the unlabeled zero-sequence current data to obtain an encoder and a decoder; A nonlinear mapping is established between the encoder, the decoder and a high-resistance grounding fault label of a preset fully connected network layer to generate a preset high-resistance grounding fault identification model.

8. The decision basis analysis device for the high-resistance grounding fault identification model according to claim 7, characterized in that: Also includes: Training function building blocks, specifically for: Calculate the fitting error between the input zero-sequence current wave and the decoded reconstructed wave; Calculate the waveform similarity between the input zero-sequence current wave and the decoded reconstructed wave; An objective function of the encoder is constructed according to the fitting error and the waveform similarity.

9. The decision basis analysis device for the high-resistance grounding fault identification model according to claim 6, characterized in that: The feature analysis unit is specifically used to: Constructing an additive explanatory model based on the Shapley value of game theory; Analyzing the importance of each encoding feature of each sample output by the encoder to the model recognition result according to the additive interpretation model to obtain a local Shapley value; The absolute average of the local Shapley values ​​corresponding to the coding features at the same position of all samples is calculated to obtain a global Shapley value.

10. Decision basis analysis equipment for high resistance ground fault identification model, characterized in that: The device includes a processor and a memory; The memory is used to store program code and transmit the program code to the processor; The processor is configured to execute the decision basis analysis method for the high-resistance grounding fault identification model according to any one of claims 1 to 5 according to the instructions in the program code.

Images