Single-phase earth fault early warning method, device, equipment and storage medium

Abstract

The invention discloses a single-phase earth fault early warning method, device and equipment and a storage medium, and relates to the technical field of electric power operation and maintenance, and the method comprises the steps: carrying out the simulation of a single-phase earth fault through a single-phase earth cabinet, and collecting a ratio difference change curve before and after the occurrence of the single-phase earth fault through a voltage transformer; determining an error matrix based on the ratio difference change curve, and calculating the matrix by using a linear kernel function, a polynomial kernel function, a Gaussian kernel function and a Sigmoid kernel function to obtain a hierarchical combination calculation matrix; and calculating the hierarchical combination calculation matrix by using the initial neural network, adjusting parameters of the initial neural network based on an obtained learning rate calculation result, fitting data corresponding to the to-be-detected power distribution network by using the obtained target neural network to obtain a fitted curve, and comparing the fitted curve with the ratio difference change curve to obtain a ratio difference change curve of the to-be-detected power distribution network. And if the comparison result shows that the information is consistent, sending an early warning signal. Therefore, the fault detection speed and accuracy of the power distribution network can be improved.

Description

Technical Field

[0001] The present application relates to the field of power operation and maintenance technology, and in particular to a single-phase grounding fault early warning method, device, equipment and storage medium. Background Art

[0002] Currently, 6kV to 35kV distribution networks typically operate with a neutral point that is not effectively grounded. If a single-phase ground fault occurs in the system—that is, a short circuit between one phase of a three-phase system and the ground—the voltage in the faulted phase will drop to near zero, while the voltage in the unfaulted phase will rise to the line voltage. Furthermore, while the fault may generate arcing, smoke, or unusual noise, it is not necessary to directly disconnect the faulty line; the system can operate with the fault for two hours. However, when the distribution network is energized during fault operation, whether the voltage transformer measurement error exceeds the standard and thus causes inaccurate energy measurement is of great concern to both power suppliers and users.

[0003] As can be seen from the above, how to improve the efficiency of fault detection in the distribution network during the single-phase grounding fault early warning process is an urgent problem to be solved. Summary of the Invention

[0004] In view of this, the purpose of this application is to provide a single-phase ground fault early warning method, device, equipment and storage medium, which can improve the speed and accuracy of fault detection in the distribution network, thereby improving the efficiency of the production process. The specific solution is as follows:

[0005] In a first aspect, the present application provides a single-phase ground fault early warning method, comprising:

[0006] The single-phase grounding fault is simulated using the single-phase grounding cabinet in the single-phase grounding fault test platform, and the voltage transformer is used to collect the ratio difference change curve before and after the single-phase grounding fault occurs;

[0007] Determine an error matrix based on the ratio difference change curve, and use a preset linear kernel function, a preset polynomial kernel function, a preset Gaussian kernel function, and a preset Sigmoid kernel function to calculate the error matrix in sequence to obtain a hierarchical combination calculation matrix;

[0008] Calculating the hierarchical combination calculation matrix using a preset learning rate calculation function in the initial neural network to obtain a learning rate calculation result, and adjusting the parameters corresponding to the initial neural network based on a preset error loss calculation function and the learning rate calculation result to obtain a target neural network;

[0009] The target neural network is used to fit the data corresponding to the distribution network to be detected to obtain a fitting curve, and the fitting curve is compared with the ratio difference change curve. If the comparison result is consistent, a warning signal corresponding to the single-phase grounding fault is generated and sent.

[0010] Optionally, the single-phase grounding fault simulation using a single-phase grounding cabinet in a single-phase grounding fault test platform includes:

[0011] A single-phase grounding fault test platform is built based on a step-up transformer, a circuit breaker, a simulation line, a single-phase grounding cabinet, a voltage transformer, a step-down transformer, and a controller; the step-up transformer is used to step up the three-phase AC voltage and transmit the stepped-up voltage to the step-down transformer via the simulation line and the single-phase grounding cabinet; the voltage transformer is located on the line behind the single-phase grounding cabinet;

[0012] After the controller receives the single-phase grounding fault simulation instruction, the controller is called and the single-phase grounding cabinet is controlled based on the single-phase grounding fault simulation instruction to implement a single-phase grounding fault simulation operation.

[0013] Optionally, the collecting of a ratio difference change curve before and after a single-phase grounding fault occurs by using a voltage transformer includes:

[0014] After the single-phase grounding fault test platform is powered on, using the voltage transformer to collect first voltage data corresponding to before the single-phase grounding fault occurs at first preset time intervals within a preset time range;

[0015] When the power-on time of the single-phase grounding fault test platform is within the first preset time range, the controller is used to control the single-phase grounding cabinet to perform a single-phase grounding fault simulation, and the voltage transformer is used to collect second voltage data corresponding to the occurrence of the single-phase grounding fault at every first preset time interval within the preset time range;

[0016] A ratio difference change curve is established based on the first voltage data and the second voltage data.

[0017] Optionally, the error matrix is determined based on the ratio difference change curve, and the error matrix is calculated in sequence using a preset linear kernel function, a preset polynomial kernel function, a preset Gaussian kernel function, and a preset Sigmoid kernel function to obtain a hierarchical combination calculation matrix, including:

[0018] Calculating the ratio difference change curve at every second preset time interval using a preset error matrix calculation rule to obtain an error matrix;

[0019] Calculating each row vector in the error matrix using a preset linear kernel function to obtain a corresponding first calculation result, and determining a linear kernel calculation matrix based on each first calculation result;

[0020] Calculating each row vector in the linear kernel calculation matrix using a preset polynomial kernel function and based on the corresponding first set of hyperparameters to obtain a corresponding second calculation result, and determining a polynomial kernel calculation matrix based on each of the second calculation results;

[0021] Calculating each row vector in the polynomial kernel calculation matrix using a preset Gaussian kernel function and based on the corresponding second set of hyperparameters to obtain a corresponding third calculation result, and determining a Gaussian kernel calculation matrix based on each of the third calculation results;

[0022] Using a preset Sigmoid kernel function and based on the corresponding third set of hyperparameters, each row vector in the Gaussian kernel calculation matrix is calculated to obtain a corresponding fourth calculation result, and a hierarchical combination calculation matrix is determined based on each of the fourth calculation results.

[0023] Optionally, after determining the error matrix based on the ratio difference change curve, and sequentially calculating the error matrix using a preset linear kernel function, a preset polynomial kernel function, a preset Gaussian kernel function, and a preset Sigmoid kernel function to obtain a hierarchical combination calculation matrix, the method further includes:

[0024] Processing the first set of hyperparameters, the second set of hyperparameters, and the third set of hyperparameters using a preset loss function and a preset regularization function to obtain corresponding new hyperparameters to be used, and processing the error matrix using the hyperparameters to be used to obtain a new hierarchical combination calculation matrix;

[0025] Performing a centralization operation on the hierarchical combination calculation matrix according to a preset centralization rule to obtain a centralized core matrix, and performing an extraction operation on the centralized core matrix using a preset eigenvalue extraction rule and a preset eigenvector extraction rule to obtain a plurality of eigenvalues and their corresponding eigenvectors;

[0026] The eigenvector corresponding to the eigenvalue with the largest value among the eigenvalues is set as the principal component, and a dimensionality reduction space is established using the principal component, so as to project the hierarchical combination calculation matrix into the dimensionality reduction space to obtain the hierarchical combination calculation matrix after dimensionality reduction.

[0027] Optionally, the layer combination calculation matrix is calculated using a preset learning rate calculation function in the initial neural network to obtain a learning rate calculation result, and the parameters corresponding to the initial neural network are adjusted based on a preset error loss calculation function and the learning rate calculation result to obtain a target neural network, including:

[0028] Determine a minimum learning rate and a maximum learning rate corresponding to the initial neural network, and determine a learning rate calculation result based on the minimum learning rate, the maximum learning rate, the current training loss corresponding to the hierarchical combination calculation matrix, and the previous training loss using a preset learning rate calculation function;

[0029] The preset error loss calculation function is used and based on the learning rate calculation result, the network parameters corresponding to the initial neural network are adjusted to obtain the target neural network.

[0030] Optionally, the step of fitting the data corresponding to the distribution network to be detected using the target neural network to obtain a fitting curve, and comparing the fitting curve with the ratio difference change curve; if the obtained comparison results are consistent, generating and sending a warning signal corresponding to a single-phase grounding fault includes:

[0031] Performing real-time monitoring on the working data corresponding to the voltage transformer in the distribution network to be detected to obtain monitoring data;

[0032] Fitting the monitoring data using the target neural network to obtain a corresponding fitting curve, and comparing a first curve change characteristic corresponding to the fitting curve with a second curve change characteristic corresponding to the ratio difference change curve to obtain a comparison result;

[0033] If the comparison result indicates that the first curve change characteristic and the second curve change characteristic have the same regularity, the state corresponding to the distribution network to be detected is set to a single-phase grounding fault state, and then a single-phase grounding fault warning signal is generated and sent.

[0034] In a second aspect, the present application provides a single-phase ground fault early warning device, comprising:

[0035] A change curve determination module is used to simulate a single-phase grounding fault using a single-phase grounding cabinet in a single-phase grounding fault test platform, and to collect a ratio difference change curve before and after the single-phase grounding fault occurs using a voltage transformer;

[0036] a matrix determination module, configured to determine an error matrix based on the ratio difference change curve, and sequentially calculate the error matrix using a preset linear kernel function, a preset polynomial kernel function, a preset Gaussian kernel function, and a preset Sigmoid kernel function to obtain a hierarchical combination calculation matrix;

[0037] a neural network determination module, configured to calculate the hierarchical combination calculation matrix using a preset learning rate calculation function in the initial neural network to obtain a learning rate calculation result, and to adjust parameters corresponding to the initial neural network based on a preset error loss calculation function and the learning rate calculation result to obtain a target neural network;

[0038] The fault detection module is used to use the target neural network to fit the data corresponding to the distribution network to be detected to obtain a fitting curve, and compare the fitting curve with the ratio difference change curve. If the comparison result is consistent, a warning signal corresponding to the single-phase grounding fault is generated and sent.

[0039] Optionally, the change curve determination module includes:

[0040] A fault test platform establishment unit is used to build a single-phase grounding fault test platform based on a step-up transformer, a circuit breaker, a simulation line, a single-phase grounding cabinet, a voltage transformer, a step-down transformer, and a controller; the step-up transformer is used to step up the three-phase AC voltage and transmit the stepped-up voltage to the step-down transformer via the simulation line and the single-phase grounding cabinet; the voltage transformer is located on the line behind the single-phase grounding cabinet;

[0041] The single-phase grounding cabinet control unit is used to call the controller after the controller receives the single-phase grounding fault simulation instruction and control the single-phase grounding cabinet based on the single-phase grounding fault simulation instruction to realize the single-phase grounding fault simulation operation.

[0042] Optionally, the change curve determination module includes:

[0043] a first voltage data determining unit configured to, after the single-phase grounding fault test platform is powered on, use the voltage transformer to collect first voltage data corresponding to before a single-phase grounding fault occurs at first preset time intervals within a preset time range;

[0044] a second voltage data determining unit, configured to, when the power-on time of the single-phase grounding fault test platform is within the first preset time range, use the controller to control the single-phase grounding cabinet to perform a single-phase grounding fault simulation, and use the voltage transformer to collect second voltage data corresponding to the occurrence of the single-phase grounding fault at every first preset time interval within the preset time range;

[0045] The change curve determining subunit is configured to establish a ratio difference change curve based on the first voltage data and the second voltage data.

[0046] Optionally, the matrix determination module includes:

[0047] an error matrix calculation unit, configured to calculate the ratio difference change curve at every second preset time interval using a preset error matrix calculation rule to obtain an error matrix;

[0048] a first calculation matrix determination unit, configured to calculate each row vector in the error matrix using a preset linear kernel function to obtain a corresponding first calculation result, and determine a linear kernel calculation matrix based on each first calculation result;

[0049] a second calculation matrix determination unit, configured to calculate each row vector in the linear kernel calculation matrix using a preset polynomial kernel function and based on the corresponding first set of hyperparameters to obtain a corresponding second calculation result, and determine a polynomial kernel calculation matrix based on each of the second calculation results;

[0050] a third calculation matrix determination unit, configured to calculate each row vector in the polynomial kernel calculation matrix using a preset Gaussian kernel function and based on a corresponding second set of hyperparameters to obtain a corresponding third calculation result, and determine a Gaussian kernel calculation matrix based on each of the third calculation results;

[0051] The matrix determination subunit is used to calculate each row vector in the Gaussian kernel calculation matrix using a preset Sigmoid kernel function and based on the corresponding third set of hyperparameters to obtain a corresponding fourth calculation result, and determine the hierarchical combination calculation matrix based on each of the fourth calculation results.

[0052] Optionally, the single-phase grounding fault early warning device further includes:

[0053] a hyperparameter determination unit, configured to process the first set of hyperparameters, the second set of hyperparameters, and the third set of hyperparameters using a preset loss function and a preset regularization function to obtain corresponding new hyperparameters to be used, and to process the error matrix using the hyperparameters to be used to obtain a new hierarchical combination calculation matrix;

[0054] a matrix extraction unit, configured to perform a centralization operation on the hierarchical combination calculation matrix according to a preset centralization rule to obtain a centralized core matrix, and perform an extraction operation on the centralized core matrix using a preset eigenvalue extraction rule and a preset eigenvector extraction rule to obtain a plurality of eigenvalues and their corresponding eigenvectors;

[0055] The eigenvector setting unit is used to set the eigenvector corresponding to the eigenvalue with the largest value among the eigenvalues as the principal component, so as to establish a dimensionality reduction space using the principal component, so as to project the hierarchical combination calculation matrix into the dimensionality reduction space to obtain the hierarchical combination calculation matrix after dimensionality reduction.

[0056] Optionally, the neural network determination module includes:

[0057] a learning rate calculation result determination unit, configured to determine a minimum learning rate and a maximum learning rate corresponding to the initial neural network, and determine a learning rate calculation result based on the minimum learning rate, the maximum learning rate, the current training loss corresponding to the hierarchical combination calculation matrix, and the previous training loss using a preset learning rate calculation function;

[0058] A neural network parameter adjustment unit is used to adjust the network parameters corresponding to the initial neural network based on the preset error loss calculation function and the learning rate calculation result to obtain a target neural network.

[0059] Optionally, the fault detection module includes:

[0060] A monitoring data determining unit, configured to perform real-time monitoring on the operating data corresponding to the voltage transformer in the distribution network to be detected to obtain monitoring data;

[0061] a comparison result determination unit, configured to fit the monitoring data using the target neural network to obtain a corresponding fitting curve, and compare a first curve change characteristic corresponding to the fitting curve with a second curve change characteristic corresponding to the ratio difference change curve to obtain a comparison result;

[0062] An early warning signal generating unit is used to set the state corresponding to the distribution network to be detected to a single-phase grounding fault state if the comparison result indicates that the first curve change characteristic and the second curve change characteristic have the same law, and then generate and send a single-phase grounding fault early warning signal.

[0063] In a third aspect, the present application provides an electronic device, comprising:

[0064] Memory, used to store computer programs;

[0065] The processor is configured to execute the computer program to implement the aforementioned single-phase grounding fault early warning method.

[0066] In a fourth aspect, the present application provides a computer-readable storage medium for storing a computer program, wherein the computer program implements the aforementioned single-phase grounding fault early warning method when executed by a processor.

[0067] As can be seen from the above, before performing a single-phase grounding fault warning, the present application needs to use the single-phase grounding cabinet in the single-phase grounding fault test platform to simulate the single-phase grounding fault, and use the voltage transformer to collect the ratio difference change curve before and after the single-phase grounding fault occurs; based on the ratio difference change curve, the error matrix is determined, and the preset linear kernel function, the preset polynomial kernel function, the preset Gaussian kernel function and the preset Sigmoid kernel function are used to calculate the error matrix in sequence to obtain a hierarchical combination calculation matrix; the preset learning rate calculation function in the initial neural network is used to calculate the hierarchical combination calculation matrix to obtain a learning rate calculation result, and the parameters corresponding to the initial neural network are adjusted based on the preset error loss calculation function and the learning rate calculation result to obtain a target neural network; the target neural network is used to fit the data corresponding to the distribution network to be detected to obtain a fitting curve, and the fitting curve is compared with the ratio difference change curve. If the comparison result obtained is consistent, a warning signal corresponding to the single-phase grounding fault is generated and sent.

[0068] It can be seen that the present application first simulates a single-phase grounding fault by using a single-phase grounding cabinet in a single-phase grounding fault test platform, and uses a voltage transformer to collect the ratio difference change curve before and after the single-phase grounding fault occurs; secondly, the error matrix is determined based on the ratio difference change curve, and the error matrix is calculated in sequence using a preset linear kernel function, a preset polynomial kernel function, a preset Gaussian kernel function, and a preset Sigmoid kernel function to obtain a hierarchical combination calculation matrix; subsequently, the hierarchical combination calculation matrix is calculated using a preset learning rate calculation function in the initial neural network to obtain a learning rate calculation result, and the parameters corresponding to the initial neural network are adjusted based on the preset error loss calculation function and the learning rate calculation result to obtain a target neural network; finally, the target neural network is used to fit the data corresponding to the distribution network to be detected to obtain a fitting curve, and the fitting curve is compared with the ratio difference change curve. If the comparison result is consistent, a warning signal corresponding to the single-phase grounding fault is generated and sent. In this way, the speed and accuracy of fault detection in the distribution network are improved, thereby improving the efficiency of the production process. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without any creative work.

[0070] Figure 1 This is a flow chart of a single-phase grounding fault early warning method disclosed in this application;

[0071] Figure 2 This is a structural diagram of a specific single-phase grounding fault test platform disclosed in this application;

[0072] Figure 3 This is a schematic diagram of a specific process disclosed in this application for sequentially processing contrast difference change curves using several kernel functions;

[0073] Figure 4 This is a structural schematic diagram of a single-phase grounding fault warning device disclosed in this application;

[0074] Figure 5 This is a structural diagram of an electronic device disclosed in this application. DETAILED DESCRIPTION

[0075] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0076] At present, the operation mode commonly adopted by the 6kV to 35kV distribution network is the neutral point non-effective grounding operation mode. If a single-phase grounding fault occurs in the system. That is, one of the phases in the three-phase system is short-circuited to the ground, the fault phase voltage will drop to near zero, while the non-fault phase voltage will rise to the line voltage. At the same time, when the fault point may produce arc light, smoke or abnormal sound, it is not necessary to directly cut off the fault line, and the fault can be operated for 2 hours. However, when the distribution network is energized in a faulty state, whether the voltage transformer metering error exceeds the standard and causes inaccurate energy metering is also highly valued by both power supply and power users. To this end, the present application provides a single-phase grounding fault early warning method, which can improve the efficiency of fault detection in the distribution network during the single-phase grounding fault early warning process.

[0077] See also Figure 1 As shown, the embodiment of the present application discloses a single-phase grounding fault early warning method, including:

[0078] Step S11: Use the single-phase grounding cabinet in the single-phase grounding fault test platform to simulate the single-phase grounding fault, and use the voltage transformer to collect the ratio difference change curve before and after the single-phase grounding fault occurs.

[0079] In this embodiment, a single-phase ground fault test platform of a 10KV distribution network is selected to collect multiple sets of voltage transformer ratio difference change data before and after a single-phase ground fault occurs using the single-phase ground fault test platform. The structural diagram of the single-phase ground fault test platform is shown in FIG. Figure 2As shown, the step-up transformer can boost the 380V three-phase AC voltage to produce a 10 kV three-phase AC power supply. It is worth noting that when the circuit breaker is closed, the electrical energy is transmitted via a simulated line to the step-down transformer, where it is then reduced to 380V for supplying the adjustable load. Furthermore, a single-phase ground fault cabinet is installed in front of the step-down transformer. Specifically, using the single-phase ground fault cabinet in the single-phase ground fault test platform to simulate a single-phase ground fault may include: building a single-phase ground fault test platform based on a step-up transformer, a circuit breaker, a simulated line, a single-phase ground fault cabinet, a voltage transformer, a step-down transformer, and a controller; the step-up transformer is used to boost the three-phase AC voltage and transmit the boosted voltage to the step-down transformer via a simulated line and the single-phase ground fault cabinet; the voltage transformer is located on the line behind the single-phase ground fault cabinet; and upon receiving a single-phase ground fault simulation command, the controller invokes the controller and controls the single-phase ground fault cabinet based on the single-phase ground fault simulation command to simulate the single-phase ground fault.

[0080] It's worth noting that before a single-phase ground fault occurs, the single-phase ground fault test platform, after a period of stable operation, will issue instructions to a controller via a computer, causing the controller to simulate the single-phase ground fault by controlling components such as contactors in the single-phase ground fault cabinet. A PT (potential transformer) is connected to the line behind the single-phase ground fault cabinet and connected to a corresponding electricity meter. Specifically, using the potential transformer to collect a ratio difference curve before and after a single-phase ground fault occurs can include: after the single-phase ground fault test platform is powered on, using the potential transformer to collect first voltage data corresponding to the occurrence of the single-phase ground fault at first preset time intervals within a preset time range; while the single-phase ground fault test platform is powered on within the first preset time range, using the controller to control the single-phase ground fault cabinet to simulate the single-phase ground fault, and using the potential transformer to collect second voltage data corresponding to the occurrence of the single-phase ground fault at first preset time intervals within a preset time range; and establishing a ratio difference curve based on the first and second voltage data.

[0081] In one specific implementation, the present invention issues a command to the controller 10 seconds after the platform is powered on. The controller then controls the contactors and other components in the single-phase grounding cabinet to simulate a single-phase grounding fault, while the system continues to operate with power. The system then operates under the single-phase grounding fault for 20 seconds before powering off. Even after the power outage, the system continues to measure the PT error of the faulty phase, recording the ratio difference curves of 10 sets of faulty phase PTs over a 30-second period.

[0082] Step S12: determining an error matrix based on the ratio difference change curve, and calculating the error matrix in sequence using a preset linear kernel function, a preset polynomial kernel function, a preset Gaussian kernel function, and a preset Sigmoid kernel function to obtain a hierarchical combination calculation matrix.

[0083] In this embodiment, after collecting the ratio difference change curve, the embodiment of the present application needs to determine an error matrix based on the ratio difference change curve. In a specific embodiment, based on the collected multiple sets of ratio difference change curves, the embodiment of the present application can collect an error point every 1 second from each ratio difference change curve to form an error matrix X with a size of 10*30.

[0084] It is worth mentioning that in the process of determining the corresponding error matrix based on the contrast difference change curve, the embodiment of the present application needs to use several kernel functions to process the contrast difference change curve in sequence to obtain the corresponding error matrix. In a specific embodiment, the embodiment of the present application respectively uses the polynomial kernel function, the Gaussian kernel function and the Sigmoid kernel function to process the contrast difference change curve in sequence. That is, linear kernel functions and polynomial kernel functions are used at the bottom layer to capture the basic linear and nonlinear relationships in the contrast difference change curve, while Gaussian kernel functions with larger bandwidth are used at the high layer to capture the global distribution characteristics of the data, and Sigmoid kernel functions are used to capture the local characteristics and detail information of the data, and the specific process is as follows. Figure 3 shown.

[0085] First, the embodiment of the present application needs to use the preset error matrix calculation rules and calculate the contrast difference change curve every second preset time interval to obtain the error matrix; use the preset linear kernel function to calculate each row vector in the error matrix to obtain the corresponding first calculation result, and determine the linear kernel calculation matrix based on each first calculation result.

[0086] In a specific embodiment, the present application embodiment uses linear check 、 The inner product of two vectors is performed and the expression is as follows:

[0087] ;

[0088] in, 、 express Any row vector in , ; Is a scalar quantity, indicating that and The calculated kernel function value, matrix is the kernel matrix after linear kernel calculation, its size is 10*10, and 、 is a matrix where is any row vector.

[0089] Furthermore, the embodiment of the present application needs to use a preset polynomial kernel function and calculate each row vector in the linear kernel calculation matrix based on the corresponding first set of hyperparameters to obtain a corresponding second calculation result, and determine the polynomial kernel calculation matrix based on each second calculation result. In a specific embodiment, the embodiment of the present application selects a polynomial kernel to process each row vector in the linear kernel calculation matrix, and the expression is as follows:

[0090] ;

[0091] in, , 、 、 is a hyperparameter, is a positive number, Non-negative, and order Not too high. Is a scalar quantity, indicating that and The calculated polynomial kernel value, It is the kernel matrix after polynomial kernel calculation, and its size is 10*10. 、 is a matrix where is any row vector.

[0092] Furthermore, the embodiment of the present application needs to use a preset Gaussian kernel function and calculate each row vector in the polynomial kernel calculation matrix based on the corresponding second set of hyperparameters to obtain a corresponding third calculation result, and determine the Gaussian kernel calculation matrix based on each third calculation result. In a specific embodiment, the embodiment of the present application selects a Gaussian kernel to process the polynomial kernel calculation matrix, and the expression is as follows:

[0093] ;

[0094] in, , is a hyperparameter, , Used to control the width of the kernel function; express and The Euclidean distance of Is a scalar quantity, indicating that and The calculated Gaussian kernel value, It is the kernel matrix after Gaussian kernel calculation, and its size is 10*10. Definition 、 is a matrix where is any row vector.

[0095] Finally, the embodiment of the present application uses a preset Sigmoid kernel function and calculates each row vector in the Gaussian kernel calculation matrix based on the corresponding third set of hyperparameters to obtain a corresponding fourth calculation result, and determines the hierarchical combination calculation matrix based on each fourth calculation result. In a specific embodiment, the embodiment of the present application uses a Sigmoid kernel to process the Gaussian kernel calculation matrix, and the expression is as follows:

[0096] ;

[0097] in, , 、 is a hyperparameter; Is a scalar quantity, indicating that and The calculated Sigmoid kernel value, It is the kernel matrix after Sigmoid kernel calculation, and is also the final output result of the hierarchical combination kernel function. Its size is 10*10.

[0098] It is worth mentioning that when the error matrix is processed in sequence using several kernel functions, the embodiment of the present application uses corresponding parameters to process the error matrix. In addition, since the selection of different parameters affects the effect of feature dimensionality reduction, in this embodiment, the hyperparameters are automatically learned by minimizing the preset objective function, and the expression is as follows:

[0099] ;

[0100] in, The loss function representing the dimensionality reduction result caused by the kernel matrix calculated with different parameters can be calculated using the reconstruction error function; It is a regularization term, which can use L1 or L2 regularization to prevent overfitting and ensure the sparsity of weights; The value can be 0.1.

[0101] Then, the parameter that minimizes the function value of the preset objective function is set as the corresponding parameter. Further, after obtaining the hierarchical combination calculation matrix, the embodiment of the present application needs to set the kernel matrix Centering is performed by columns or rows to obtain a centralized kernel matrix, and the eigenvalues and eigenvectors of the covariance matrix of the centralized kernel matrix are solved. The eigenvector corresponding to the largest eigenvalue is selected from the obtained calculation results as the principal component to construct a new dimensionality reduction space. Finally, the original data set is projected into the new dimensionality reduction space to obtain the reduced dimensionality data matrix. Specifically, the error matrix is determined based on the ratio difference change curve, and the error matrix is calculated in sequence using a preset linear kernel function, a preset polynomial kernel function, a preset Gaussian kernel function, and a preset Sigmoid kernel function to obtain the hierarchical combination calculation matrix. It can also include: using a preset loss function and a preset regularization function to process the first set of hyperparameters, the second set of hyperparameters, and the third set of hyperparameters to obtain corresponding new hyperparameters to be used, and using the hyperparameters to be used to process the error matrix to obtain a new hierarchical combination calculation matrix; centering the hierarchical combination calculation matrix according to a preset centering rule to obtain a centralized kernel matrix, and extracting the centralized kernel matrix using a preset eigenvalue extraction rule and a preset eigenvector extraction rule to obtain several eigenvalues and their corresponding eigenvectors; setting the eigenvector corresponding to the eigenvalue with the largest value in each eigenvalue as the principal component, and using the principal component to establish a dimensionality reduction space, and projecting the hierarchical combination calculation matrix to the dimensionality reduction space to obtain the hierarchical combination calculation matrix after dimensionality reduction.

[0102] Step S13: Calculate the hierarchical combination calculation matrix using the preset learning rate calculation function in the initial neural network to obtain a learning rate calculation result, and adjust the parameters corresponding to the initial neural network based on the preset error loss calculation function and the learning rate calculation result to obtain a target neural network.

[0103] In this embodiment, after obtaining the hierarchical combination calculation matrix after dimensionality reduction, the embodiment of the present application needs to use a BP neural network to fit the hierarchical combination calculation matrix after dimensionality reduction to learn the change pattern of the voltage transformer ratio difference before and after the single-phase grounding fault occurs. In addition, in the process of fitting the hierarchical combination calculation matrix after dimensionality reduction using the BP neural network, the embodiment of the present application can use the back propagation algorithm to adjust the weights and biases in the network so that the output of the network is as close to the true label or target value as possible. Specifically, the hierarchical combination calculation matrix is calculated using the preset learning rate calculation function in the initial neural network to obtain a learning rate calculation result, and the parameters corresponding to the initial neural network are adjusted based on the preset error loss calculation function and the learning rate calculation result to obtain the target neural network, which may include: determining the minimum learning rate and maximum learning rate corresponding to the initial neural network, and using the preset learning rate calculation function and based on the minimum learning rate, maximum learning rate, the current training loss corresponding to the hierarchical combination calculation matrix and the previous training loss to determine the learning rate calculation result; using the preset error loss calculation function and based on the learning rate calculation result to adjust the network parameters corresponding to the initial neural network to obtain the target neural network. In a specific implementation, the embodiment of the present application uses a mean square error loss function to train the BP fitting neural network during the process of establishing the BP fitting neural network, and ends the training when the loss function value is less than 0.05.

[0104] It is worth mentioning that if the learning rate of the BP neural network is changes between In a specific embodiment, is the training loss corresponding to the nth epoch, is the training loss corresponding to the previous epoch, then:

[0105] ;

[0106] in, Is a positive number used to adjust the impact of loss changes on the learning rate; is the learning rate of the nth epoch; when the loss function of the nth epoch changes significantly compared to the previous round, The value of is close to 1, so Close to the maximum value q; on the contrary, when the loss function of the nth epoch changes less than that of the previous round, The value of is close to 0, so is close to the minimum value p.

[0107] Furthermore, after the model fitting is completed, the embodiment of the present application needs to observe and analyze the changing characteristics of the fitting curve, and set the obtained changing characteristics as the ratio difference changing characteristics of the voltage transformer under a single-phase grounding fault. That is, the working state of the voltage transformer is monitored in real time to determine whether the actual ratio difference curve and the fitting curve have the same pattern. If they have the same pattern, an early warning signal can be generated and issued. Specifically, the target neural network is used to fit the data corresponding to the distribution network to be detected to obtain a fitting curve, and the fitting curve is compared with the ratio difference changing curve. If the comparison result shows consistency, an early warning signal corresponding to the single-phase grounding fault is generated and issued. This may include: real-time monitoring of the working data corresponding to the voltage transformer in the distribution network to be detected to obtain monitoring data; fitting the monitoring data using the target neural network to obtain a corresponding fitting curve, and comparing the first curve changing characteristics corresponding to the fitting curve with the second curve changing characteristics corresponding to the ratio difference changing curve to obtain a comparison result; if the comparison result shows that the first curve changing characteristics and the second curve changing characteristics have the same pattern, the state corresponding to the distribution network to be detected is set as a single-phase grounding fault state, and then a single-phase grounding fault early warning signal is generated and issued.

[0108] Thus, the embodiment of the present application first simulates a single-phase grounding fault by using a single-phase grounding cabinet in a single-phase grounding fault test platform, and uses a voltage transformer to collect a ratio difference change curve before and after the single-phase grounding fault occurs; secondly, an error matrix is determined based on the ratio difference change curve, and the error matrix is calculated in sequence using a preset linear kernel function, a preset polynomial kernel function, a preset Gaussian kernel function, and a preset Sigmoid kernel function to obtain a hierarchical combination calculation matrix; subsequently, the hierarchical combination calculation matrix is calculated using a preset learning rate calculation function in the initial neural network to obtain a learning rate calculation result, and the parameters corresponding to the initial neural network are adjusted based on the preset error loss calculation function and the learning rate calculation result to obtain a target neural network; finally, the target neural network is used to fit the data corresponding to the distribution network to be detected to obtain a fitting curve, and the fitting curve is compared with the ratio difference change curve. If the comparison result is consistent, a warning signal corresponding to the single-phase grounding fault is generated and sent. In this way, the speed and accuracy of fault detection in the distribution network are improved, thereby improving the efficiency of the production process.

[0109] Accordingly, see Figure 4 As shown, the present application also provides a single-phase grounding fault early warning device, comprising:

[0110] The change curve determination module 11 is used to simulate a single-phase grounding fault using a single-phase grounding cabinet in a single-phase grounding fault test platform, and to collect a ratio difference change curve before and after the single-phase grounding fault occurs using a voltage transformer;

[0111] a matrix determination module 12 for determining an error matrix based on the ratio difference change curve, and calculating the error matrix in sequence using a preset linear kernel function, a preset polynomial kernel function, a preset Gaussian kernel function, and a preset Sigmoid kernel function to obtain a hierarchical combination calculation matrix;

[0112] A neural network determination module 13 is configured to calculate the hierarchical combination calculation matrix using a preset learning rate calculation function in the initial neural network to obtain a learning rate calculation result, and to adjust the parameters corresponding to the initial neural network based on a preset error loss calculation function and the learning rate calculation result to obtain a target neural network;

[0113] The fault detection module 14 is used to use the target neural network to fit the data corresponding to the distribution network to be detected to obtain a fitting curve, and compare the fitting curve with the ratio difference change curve. If the comparison result is consistent, a warning signal corresponding to the single-phase grounding fault is generated and sent.

[0114] As can be seen from the above, before performing a single-phase grounding fault warning, the embodiment of the present application first simulates a single-phase grounding fault by using a single-phase grounding cabinet in a single-phase grounding fault test platform, and uses a voltage transformer to collect a ratio difference change curve before and after the single-phase grounding fault occurs; secondly, an error matrix is determined based on the ratio difference change curve, and the error matrix is calculated in sequence using a preset linear kernel function, a preset polynomial kernel function, a preset Gaussian kernel function, and a preset Sigmoid kernel function to obtain a hierarchical combination calculation matrix; subsequently, the hierarchical combination calculation matrix is calculated using a preset learning rate calculation function in the initial neural network to obtain a learning rate calculation result, and the parameters corresponding to the initial neural network are adjusted based on the preset error loss calculation function and the learning rate calculation result to obtain a target neural network; finally, the target neural network is used to fit the data corresponding to the distribution network to be detected to obtain a fitting curve, and the fitting curve is compared with the ratio difference change curve. If the comparison result is consistent, a warning signal corresponding to the single-phase grounding fault is generated and sent. In this way, the speed and accuracy of fault detection in the distribution network are improved, thereby improving the efficiency of the production process.

[0115] In some specific implementations, the change curve determination module 11 may specifically include:

[0116] A fault test platform establishment unit is used to build a single-phase grounding fault test platform based on a step-up transformer, a circuit breaker, a simulation line, a single-phase grounding cabinet, a voltage transformer, a step-down transformer, and a controller; the step-up transformer is used to step up the three-phase AC voltage and transmit the stepped-up voltage to the step-down transformer via the simulation line and the single-phase grounding cabinet; the voltage transformer is located on the line behind the single-phase grounding cabinet;

[0117] The single-phase grounding cabinet control unit is used to call the controller after the controller receives the single-phase grounding fault simulation instruction and control the single-phase grounding cabinet based on the single-phase grounding fault simulation instruction to realize the single-phase grounding fault simulation operation.

[0118] In some specific implementations, the change curve determination module 11 may specifically include:

[0119] a first voltage data determining unit configured to, after the single-phase grounding fault test platform is powered on, use the voltage transformer to collect first voltage data corresponding to before a single-phase grounding fault occurs at first preset time intervals within a preset time range;

[0120] a second voltage data determining unit, configured to, when the power-on time of the single-phase grounding fault test platform is within the first preset time range, use the controller to control the single-phase grounding cabinet to perform a single-phase grounding fault simulation, and use the voltage transformer to collect second voltage data corresponding to the occurrence of the single-phase grounding fault at every first preset time interval within the preset time range;

[0121] The change curve determining subunit is configured to establish a ratio difference change curve based on the first voltage data and the second voltage data.

[0122] In some specific implementations, the matrix determination module 12 may specifically include:

[0123] an error matrix calculation unit, configured to calculate the ratio difference change curve at every second preset time interval using a preset error matrix calculation rule to obtain an error matrix;

[0124] a first calculation matrix determination unit, configured to calculate each row vector in the error matrix using a preset linear kernel function to obtain a corresponding first calculation result, and determine a linear kernel calculation matrix based on each first calculation result;

[0125] a second calculation matrix determination unit, configured to calculate each row vector in the linear kernel calculation matrix using a preset polynomial kernel function and based on the corresponding first set of hyperparameters to obtain a corresponding second calculation result, and determine a polynomial kernel calculation matrix based on each of the second calculation results;

[0126] a third calculation matrix determination unit, configured to calculate each row vector in the polynomial kernel calculation matrix using a preset Gaussian kernel function and based on a corresponding second set of hyperparameters to obtain a corresponding third calculation result, and determine a Gaussian kernel calculation matrix based on each of the third calculation results;

[0127] The matrix determination subunit is used to calculate each row vector in the Gaussian kernel calculation matrix using a preset Sigmoid kernel function and based on the corresponding third set of hyperparameters to obtain a corresponding fourth calculation result, and determine the hierarchical combination calculation matrix based on each of the fourth calculation results.

[0128] In some specific implementations, the single-phase grounding fault warning device may further include:

[0129] a hyperparameter determination unit, configured to process the first set of hyperparameters, the second set of hyperparameters, and the third set of hyperparameters using a preset loss function and a preset regularization function to obtain corresponding new hyperparameters to be used, and to process the error matrix using the hyperparameters to be used to obtain a new hierarchical combination calculation matrix;

[0130] a matrix extraction unit, configured to perform a centralization operation on the hierarchical combination calculation matrix according to a preset centralization rule to obtain a centralized core matrix, and perform an extraction operation on the centralized core matrix using a preset eigenvalue extraction rule and a preset eigenvector extraction rule to obtain a plurality of eigenvalues and their corresponding eigenvectors;

[0131] The eigenvector setting unit is used to set the eigenvector corresponding to the eigenvalue with the largest value among the eigenvalues as the principal component, so as to establish a dimensionality reduction space using the principal component, so as to project the hierarchical combination calculation matrix into the dimensionality reduction space to obtain the hierarchical combination calculation matrix after dimensionality reduction.

[0132] In some specific implementations, the neural network determination module 13 may specifically include:

[0133] a learning rate calculation result determination unit, configured to determine a minimum learning rate and a maximum learning rate corresponding to the initial neural network, and determine a learning rate calculation result based on the minimum learning rate, the maximum learning rate, the current training loss corresponding to the hierarchical combination calculation matrix, and the previous training loss using a preset learning rate calculation function;

[0134] A neural network parameter adjustment unit is used to adjust the network parameters corresponding to the initial neural network based on the preset error loss calculation function and the learning rate calculation result to obtain a target neural network.

[0135] In some specific implementations, the fault detection module 14 may specifically include:

[0136] A monitoring data determining unit, configured to perform real-time monitoring on the operating data corresponding to the voltage transformer in the distribution network to be detected to obtain monitoring data;

[0137] a comparison result determination unit, configured to fit the monitoring data using the target neural network to obtain a corresponding fitting curve, and compare a first curve change characteristic corresponding to the fitting curve with a second curve change characteristic corresponding to the ratio difference change curve to obtain a comparison result;

[0138] An early warning signal generating unit is used to set the state corresponding to the distribution network to be detected to a single-phase grounding fault state if the comparison result indicates that the first curve change characteristic and the second curve change characteristic have the same law, and then generate and send a single-phase grounding fault early warning signal.

[0139] Furthermore, the embodiment of the present application also discloses an electronic device, Figure 5 This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content of this diagram should not be construed as limiting the scope of application of this application. The electronic device 20 may specifically include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 is used to store a computer program, which is loaded and executed by the processor 21 to implement the relevant steps of the single-phase ground fault early warning method disclosed in any of the aforementioned embodiments. Furthermore, the electronic device 20 in this embodiment may specifically be an electronic computer.

[0140] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and the external device. The communication protocol it follows is any communication protocol that can be applied to the technical solution of this application and is not specifically limited here; the input and output interface 25 is used to obtain external input data or output data to the outside world. Its specific interface type can be selected according to specific application needs and is not specifically limited here.

[0141] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or CD, etc. The resources stored thereon can include an operating system 221, a computer program 222, etc., and the storage method can be temporary storage or permanent storage.

[0142] The operating system 221 is used to manage and control the hardware devices on the electronic device 20 and the computer program 222. The operating system 221 can be Windows Server, NetWare, Unix, Linux, etc. In addition to including a computer program capable of implementing the single-phase grounding fault early warning method performed by the electronic device 20 as disclosed in any of the aforementioned embodiments, the computer program 222 can further include a computer program capable of implementing other specific tasks.

[0143] Furthermore, this application discloses a computer-readable storage medium for storing a computer program. When executed by a processor, the computer program implements the single-phase ground fault early warning method disclosed above. The specific steps of this method can be found in the corresponding contents disclosed in the aforementioned embodiments and will not be further described here.

[0144] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from the other embodiments. Reference can be made to the descriptions of the identical or similar parts between the various embodiments. For the devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple, and the relevant parts can be referred to the descriptions of the methods.

[0145] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the components and steps of each example according to their functions. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0146] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in random access memory (RAM), internal memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0147] Finally, it should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. 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 device comprising the element.

[0148] The above is a detailed introduction to the technical solution provided by the present application. Specific examples are used herein to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea. At the same time, for those skilled in the art, according to the ideas of the present application, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A single-phase ground fault early warning method, characterized in that: include: The single-phase grounding fault is simulated using the single-phase grounding cabinet in the single-phase grounding fault test platform, and the voltage transformer is used to collect the ratio difference change curve before and after the single-phase grounding fault occurs; Determine an error matrix based on the ratio difference change curve, and use a preset linear kernel function, a preset polynomial kernel function, a preset Gaussian kernel function, and a preset Sigmoid kernel function to calculate the error matrix in sequence to obtain a hierarchical combination calculation matrix; Calculating the hierarchical combination calculation matrix using a preset learning rate calculation function in the initial neural network to obtain a learning rate calculation result, and adjusting the parameters corresponding to the initial neural network based on a preset error loss calculation function and the learning rate calculation result to obtain a target neural network; The target neural network is used to fit the data corresponding to the distribution network to be detected to obtain a fitting curve, and the fitting curve is compared with the ratio difference change curve. If the comparison result is consistent, a warning signal corresponding to the single-phase grounding fault is generated and sent.

2. The single-phase grounding fault early warning method according to claim 1, characterized in that: The single-phase grounding fault simulation using the single-phase grounding cabinet in the single-phase grounding fault test platform includes: A single-phase grounding fault test platform is built based on a step-up transformer, a circuit breaker, a simulation line, a single-phase grounding cabinet, a voltage transformer, a step-down transformer, and a controller; the step-up transformer is used to step up the three-phase AC voltage and transmit the stepped-up voltage to the step-down transformer via the simulation line and the single-phase grounding cabinet; the voltage transformer is located on the line behind the single-phase grounding cabinet; After the controller receives the single-phase grounding fault simulation instruction, the controller is called and the single-phase grounding cabinet is controlled based on the single-phase grounding fault simulation instruction to implement a single-phase grounding fault simulation operation.

3. The single-phase grounding fault early warning method according to claim 2, characterized in that: The method of collecting the ratio difference change curve before and after the single-phase grounding fault occurs by using a voltage transformer includes: After the single-phase grounding fault test platform is powered on, using the voltage transformer to collect first voltage data corresponding to before the single-phase grounding fault occurs at first preset time intervals within a preset time range; When the power-on time of the single-phase grounding fault test platform is within the first preset time range, the controller is used to control the single-phase grounding cabinet to perform a single-phase grounding fault simulation, and the voltage transformer is used to collect second voltage data corresponding to the occurrence of the single-phase grounding fault at every first preset time interval within the preset time range; A ratio difference change curve is established based on the first voltage data and the second voltage data.

4. The single-phase grounding fault early warning method according to claim 1, characterized in that: The error matrix is determined based on the ratio difference change curve, and the error matrix is calculated in sequence using a preset linear kernel function, a preset polynomial kernel function, a preset Gaussian kernel function, and a preset Sigmoid kernel function to obtain a hierarchical combination calculation matrix, including: Calculating the ratio difference change curve at every second preset time interval using a preset error matrix calculation rule to obtain an error matrix; Calculating each row vector in the error matrix using a preset linear kernel function to obtain a corresponding first calculation result, and determining a linear kernel calculation matrix based on each first calculation result; Calculating each row vector in the linear kernel calculation matrix using a preset polynomial kernel function and based on the corresponding first set of hyperparameters to obtain a corresponding second calculation result, and determining a polynomial kernel calculation matrix based on each of the second calculation results; Calculating each row vector in the polynomial kernel calculation matrix using a preset Gaussian kernel function and based on the corresponding second set of hyperparameters to obtain a corresponding third calculation result, and determining a Gaussian kernel calculation matrix based on each of the third calculation results; Using a preset Sigmoid kernel function and based on the corresponding third set of hyperparameters, each row vector in the Gaussian kernel calculation matrix is calculated to obtain a corresponding fourth calculation result, and a hierarchical combination calculation matrix is determined based on each of the fourth calculation results.

5. The single-phase grounding fault early warning method according to claim 4, characterized in that: The error matrix is determined based on the ratio difference change curve, and the error matrix is calculated in sequence using a preset linear kernel function, a preset polynomial kernel function, a preset Gaussian kernel function, and a preset Sigmoid kernel function to obtain a hierarchical combination calculation matrix, further comprising: Processing the first set of hyperparameters, the second set of hyperparameters, and the third set of hyperparameters using a preset loss function and a preset regularization function to obtain corresponding new hyperparameters to be used, and processing the error matrix using the hyperparameters to be used to obtain a new hierarchical combination calculation matrix; Performing a centralization operation on the hierarchical combination calculation matrix according to a preset centralization rule to obtain a centralized core matrix, and performing an extraction operation on the centralized core matrix using a preset eigenvalue extraction rule and a preset eigenvector extraction rule to obtain a plurality of eigenvalues and their corresponding eigenvectors; The eigenvector corresponding to the eigenvalue with the largest value among the eigenvalues is set as the principal component, and a dimensionality reduction space is established using the principal component, so as to project the hierarchical combination calculation matrix into the dimensionality reduction space to obtain the hierarchical combination calculation matrix after dimensionality reduction.

6. The single-phase grounding fault early warning method according to claim 5, characterized in that: The method of calculating the hierarchical combination calculation matrix using a preset learning rate calculation function in the initial neural network to obtain a learning rate calculation result, and adjusting the parameters corresponding to the initial neural network based on a preset error loss calculation function and the learning rate calculation result to obtain a target neural network includes: Determine a minimum learning rate and a maximum learning rate corresponding to the initial neural network, and determine a learning rate calculation result based on the minimum learning rate, the maximum learning rate, the current training loss corresponding to the hierarchical combination calculation matrix, and the previous training loss using a preset learning rate calculation function; The preset error loss calculation function is used and based on the learning rate calculation result, the network parameters corresponding to the initial neural network are adjusted to obtain the target neural network.

7. The single-phase grounding fault early warning method according to any one of claims 1 to 6, characterized in that: The method of fitting the data corresponding to the distribution network to be detected by using the target neural network to obtain a fitting curve, and comparing the fitting curve with the ratio difference change curve, and generating and sending a warning signal corresponding to a single-phase grounding fault if the obtained comparison results are consistent, includes: Performing real-time monitoring on the working data corresponding to the voltage transformer in the distribution network to be detected to obtain monitoring data; Fitting the monitoring data using the target neural network to obtain a corresponding fitting curve, and comparing a first curve change characteristic corresponding to the fitting curve with a second curve change characteristic corresponding to the ratio difference change curve to obtain a comparison result; If the comparison result indicates that the first curve change characteristic and the second curve change characteristic have the same regularity, the state corresponding to the distribution network to be detected is set to a single-phase grounding fault state, and then a single-phase grounding fault warning signal is generated and sent.

8. A single-phase ground fault warning device, characterized in that: include: A change curve determination module is used to simulate a single-phase grounding fault using a single-phase grounding cabinet in a single-phase grounding fault test platform, and to collect a ratio difference change curve before and after the single-phase grounding fault occurs using a voltage transformer; a matrix determination module, configured to determine an error matrix based on the ratio difference change curve, and sequentially calculate the error matrix using a preset linear kernel function, a preset polynomial kernel function, a preset Gaussian kernel function, and a preset Sigmoid kernel function to obtain a hierarchical combination calculation matrix; a neural network determination module, configured to calculate the hierarchical combination calculation matrix using a preset learning rate calculation function in the initial neural network to obtain a learning rate calculation result, and to adjust parameters corresponding to the initial neural network based on a preset error loss calculation function and the learning rate calculation result to obtain a target neural network; The fault detection module is used to use the target neural network to fit the data corresponding to the distribution network to be detected to obtain a fitting curve, and compare the fitting curve with the ratio difference change curve. If the comparison result is consistent, a warning signal corresponding to the single-phase grounding fault is generated and sent.

9. The single-phase grounding fault early warning device according to claim 8, characterized in that: The change curve determination module includes: A fault test platform establishment unit is used to build a single-phase grounding fault test platform based on a step-up transformer, a circuit breaker, a simulation line, a single-phase grounding cabinet, a voltage transformer, a step-down transformer, and a controller; the step-up transformer is used to step up the three-phase AC voltage and transmit the stepped-up voltage to the step-down transformer via the simulation line and the single-phase grounding cabinet; the voltage transformer is located on the line behind the single-phase grounding cabinet; The single-phase grounding cabinet control unit is used to call the controller after the controller receives the single-phase grounding fault simulation instruction and control the single-phase grounding cabinet based on the single-phase grounding fault simulation instruction to realize the single-phase grounding fault simulation operation.

10. The single-phase grounding fault early warning device according to claim 9, characterized in that: The change curve determination module includes: a first voltage data determining unit configured to, after the single-phase grounding fault test platform is powered on, use the voltage transformer to collect first voltage data corresponding to before a single-phase grounding fault occurs at first preset time intervals within a preset time range; a second voltage data determining unit, configured to, when the power-on time of the single-phase grounding fault test platform is within the first preset time range, use the controller to control the single-phase grounding cabinet to perform a single-phase grounding fault simulation, and use the voltage transformer to collect second voltage data corresponding to the occurrence of the single-phase grounding fault at every first preset time interval within the preset time range; The change curve determining subunit is configured to establish a ratio difference change curve based on the first voltage data and the second voltage data.

11. The single-phase grounding fault early warning device according to claim 8, characterized in that: The matrix determination module includes: an error matrix calculation unit, configured to calculate the ratio difference change curve at every second preset time interval using a preset error matrix calculation rule to obtain an error matrix; a first calculation matrix determination unit, configured to calculate each row vector in the error matrix using a preset linear kernel function to obtain a corresponding first calculation result, and determine a linear kernel calculation matrix based on each first calculation result; a second calculation matrix determination unit, configured to calculate each row vector in the linear kernel calculation matrix using a preset polynomial kernel function and based on the corresponding first set of hyperparameters to obtain a corresponding second calculation result, and determine a polynomial kernel calculation matrix based on each of the second calculation results; a third calculation matrix determination unit, configured to calculate each row vector in the polynomial kernel calculation matrix using a preset Gaussian kernel function and based on a corresponding second set of hyperparameters to obtain a corresponding third calculation result, and determine a Gaussian kernel calculation matrix based on each of the third calculation results; The matrix determination subunit is used to calculate each row vector in the Gaussian kernel calculation matrix using a preset Sigmoid kernel function and based on the corresponding third set of hyperparameters to obtain a corresponding fourth calculation result, and determine the hierarchical combination calculation matrix based on each of the fourth calculation results.

12. The single-phase grounding fault early warning device according to claim 11, characterized in that: Also includes: a hyperparameter determination unit, configured to process the first set of hyperparameters, the second set of hyperparameters, and the third set of hyperparameters using a preset loss function and a preset regularization function to obtain corresponding new hyperparameters to be used, and to process the error matrix using the hyperparameters to be used to obtain a new hierarchical combination calculation matrix; a matrix extraction unit, configured to perform a centralization operation on the hierarchical combination calculation matrix according to a preset centralization rule to obtain a centralized core matrix, and perform an extraction operation on the centralized core matrix using a preset eigenvalue extraction rule and a preset eigenvector extraction rule to obtain a plurality of eigenvalues and their corresponding eigenvectors; The eigenvector setting unit is used to set the eigenvector corresponding to the eigenvalue with the largest value among the eigenvalues as the principal component, so as to establish a dimensionality reduction space using the principal component, so as to project the hierarchical combination calculation matrix into the dimensionality reduction space to obtain the hierarchical combination calculation matrix after dimensionality reduction.

13. The single-phase grounding fault early warning device according to claim 12, characterized in that: The neural network determination module includes: a learning rate calculation result determination unit, configured to determine a minimum learning rate and a maximum learning rate corresponding to the initial neural network, and determine a learning rate calculation result based on the minimum learning rate, the maximum learning rate, the current training loss corresponding to the hierarchical combination calculation matrix, and the previous training loss using a preset learning rate calculation function; A neural network parameter adjustment unit is used to adjust the network parameters corresponding to the initial neural network based on the preset error loss calculation function and the learning rate calculation result to obtain a target neural network.

14. The single-phase grounding fault early warning device according to any one of claims 8 to 13, characterized in that: The fault detection module includes: A monitoring data determining unit, configured to perform real-time monitoring on the operating data corresponding to the voltage transformer in the distribution network to be detected to obtain monitoring data; a comparison result determination unit, configured to fit the monitoring data using the target neural network to obtain a corresponding fitting curve, and compare a first curve change characteristic corresponding to the fitting curve with a second curve change characteristic corresponding to the ratio difference change curve to obtain a comparison result; An early warning signal generating unit is used to set the state corresponding to the distribution network to be detected to a single-phase grounding fault state if the comparison result indicates that the first curve change characteristic and the second curve change characteristic have the same law, and then generate and send a single-phase grounding fault early warning signal.

15. An electronic device, characterized in that: include: Memory, used to store computer programs; A processor is used to execute the computer program to implement the single-phase grounding fault early warning method according to any one of claims 1 to 7.

16. A computer-readable storage medium, characterized in that Used to store a computer program, wherein when the computer program is executed by a processor, the single-phase grounding fault early warning method according to any one of claims 1 to 7 is implemented.

Images