Electric shock protection method, system and equipment for low-voltage power distribution network and medium

Abstract

The invention discloses an electric shock protection method for a low-voltage power distribution network, and belongs to the field of electric shock protection of low-voltage power distribution networks. Decomposing the residual current signal into IMF components based on VMD; total energy of IMF components is calculated, and energy values of different frequency bands are used as features; inputting the features into a pre-constructed electric leakage fault type identification model, wherein the electric leakage fault type identification model identifies different electric leakage scenes; and based on the electric leakage scene identification result, executing a protection criterion. According to the method, the electric shock current component is accurately extracted from the complex residual current waveform, the degree of an electric shock accident is reflected, after the electric shock current is obtained, different action measures are executed according to the effective value of the current, and RCD protection misoperation caused by interference factors and refusal action in the electric shock state are effectively prevented.

Description

Technical Field

[0001] The present invention belongs to the field of low-voltage power grid electric shock protection, and specifically relates to a method, system, equipment and medium for identifying leakage faults and extracting electric shock current in a low-voltage distribution network containing a photovoltaic power source. Background Art

[0002] Traditional residual current protection devices (RCD, also known as leakage protection devices) use the magnitude of the residual current exceeding the threshold as the basis for determining whether the protection is activated. When the detection link detects that the residual current reaches or exceeds the threshold, the protection module can automatically disconnect the mechanical switch or combination electrical appliance of the circuit to achieve fault isolation. RCD is the most basic and important protection method for leakage faults in low-voltage distribution networks. It can effectively avoid some personal and equipment safety accidents, and electricity safety has been significantly improved. However, with the continuous transformation and upgrading of low-voltage distribution networks and the large-scale access of distributed photovoltaics, the causes of residual current have become more complex, and its value has increased accordingly. Traditional RCDs cannot effectively identify the type and source of residual current, and often malfunction, which leads to a low commissioning rate of traditional RCDs in some areas or even their inability to be put into operation.

[0003] Distributed photovoltaic power generation devices mostly use non-isolated grid-connected inverters. The non-isolated structure allows direct electrical connection between photovoltaic modules and AC power grid. The high-frequency operation of the inverter switch tube causes high-frequency common-mode voltage to be generated between the photovoltaic parasitic capacitance to the ground, the inverter, the AC power grid and the neutral point of the system, which in turn generates high-frequency common-mode leakage current, which is manifested as residual current in the AC-side RCD. When a human body is electrocuted in a traditional low-voltage distribution network without distributed photovoltaics, the electric shock current is the power frequency current, and the traditional RCD operates according to the effective value of the residual current. When connected to distributed photovoltaics, the RCD will monitor the high-frequency common-mode leakage current generated by the photovoltaic module. At this time, the waveform of the residual current is often very complex, and the increase in medium and high frequency components will cause the monitored residual current effective value to increase to varying degrees. Traditional RCDs based on current threshold detection will have the problem of false operation in low-voltage distribution networks with distributed photovoltaic access.

[0004] Existing leakage protection devices for photovoltaic leakage current are usually installed on the AC side of the photovoltaic inverter to detect the effective value of continuous photovoltaic leakage current. According to the national energy standard NB / T32004-2013 standard 7.10.2, if the continuous photovoltaic leakage current exceeds the following limit, the leakage protection device in the photovoltaic inverter should trip within 0.3s to disconnect the photovoltaic circuit and send a fault alarm signal.

[0005] 1) For PV inverters with rated output ≤30kVA, the action threshold is 300mA.

[0006] 2) Applicable to photovoltaic inverters with rated output > 30kVA, the action threshold is 10mA / kVA.

[0007] Taking a photovoltaic inverter with a rated output power of 10kVA as an example, when the effective value of the photovoltaic leakage current is 80mA, the magnitude of the leakage current does not reach the threshold of the photovoltaic inverter leakage protection device, so it will not act to cut off the distributed photovoltaic equipment. However, after the leakage current flows into the low-voltage AC grid, if its value exceeds the action threshold of the existing RCD, it will cause the protection device to malfunction, affecting the normal power consumption of users and the reliable operation of the distribution network.

[0008] When distributed photovoltaics are connected to the low-voltage distribution network on a large scale, the abnormal photovoltaic leakage current of the distributed photovoltaic modules entering the AC side will not only affect the quality of electric energy, but may also cause harm to people and property if the protection measures are insufficient or inappropriate.

[0009] The Chinese invention with the publication number CN117638788A provides a method and system for judging the leakage protection action of photovoltaic access distribution network. The statistical characteristic quantity of the leakage current is used to judge whether the collected leakage current is an abnormal leakage current. When the harmonic component of the abnormal leakage current has an energy mutation, it is judged that an electric shock fault has occurred, and the abnormal leakage current is a human body electric shock leakage current, so that the leakage current protection is activated and the fault is manually cleared; when it is judged that no electric shock fault has occurred, when the abnormal leakage current is not greater than the leakage current protection action standard value, it is determined to be a normal non-electric shock leakage current, and the leakage current protection does not act; when the abnormal leakage current is greater than the leakage current protection action standard value, it is determined to be an abnormal non-electric shock leakage current, and the abnormal non-electric shock leakage current is processed so that the effective value of the abnormal non-electric shock leakage current is less than the leakage current protection action standard value, and the leakage current protection does not act. However, this method only calculates various statistical characteristic quantities of the leakage current, and compares the collected waveform database to determine the leakage current type, which has a large amount of calculation and a slow calculation process. Summary of the invention

[0010] The object of the present invention is to provide a method, system, device and medium for electric shock protection of a low-voltage distribution network, so as to solve the problem of low efficiency in identifying leakage faults in the background technology.

[0011] In order to achieve the above object, the present invention adopts the following technical solution: In a first aspect, the present invention provides a method for protecting against electric shock in a low-voltage distribution network, comprising the following specific steps: Collect residual current signal on the user side; Decomposing the residual current signal into IMF components based on VMD; Calculate the total energy of the IMF components and use the total energy of the IMF components as a feature; Inputting the features into a pre-built leakage fault type identification model, wherein the leakage fault type identification model identifies different leakage scenarios; Execute protection criteria based on the leakage scenario identification results; Among them, when the recognition result is a normal operating state, the residual current signal continues to be collected and the protection does not operate; when the recognition result is a photovoltaic abnormal leakage, the photovoltaic safety protection criterion is executed; when the recognition result is an electric shock fault, the pre-built electric shock current extraction model is used to extract the electric shock current in the residual current signal, and the electric shock safety protection criterion is executed based on the electric shock current.

[0012] Optionally, the total energy of the IMF components is calculated and used as a feature, specifically including: Calculate the total energy of each IMF component : (1) In the formula, is the IMF component Length; Normalize the total energy of the IMF component: (2) The energy of the IMF component is used as the element to construct the feature vector as the feature of the residual current signal: .

[0013] Optionally, in the step of inputting the features into a leakage fault type identification model, the training method of the leakage fault type identification model is as follows: Obtain training data set: The data set contains three categories of data: data in normal operation, data when photovoltaic power leakage occurs, and data when electric shock occurs, corresponding to different scenarios; Construct the initial model of support vector machine SVM: construct Lagrangian function and nonlinear kernel function , set the hyperparameter combination ( , )’s initial range; The initial model of the support vector machine (SVM) is trained according to the training data set. During the training process, the sparrow search algorithm (SSA) is used to optimize the hyperparameter combination ( , ) for optimization; After the training is completed, the optimal support vector machine (SVM) model is obtained as the optimal leakage fault type identification model.

[0014] Optionally, use the SSA algorithm to optimize the hyperparameter combination ( , ) is optimized, the specific steps are: Initialize the sparrow search algorithm SSA hyperparameter combination, calculate the fitness of the current hyperparameter combination, and calculate the accuracy of the cross-validation of the leakage fault type identification model under the current hyperparameter combination; Updates the positions of discoverers, joiners, and scouts; Recalculate the fitness and determine the optimal value of the fitness; determine whether the accuracy of the cross-validation of the leakage fault type identification model meets the preset conditions under the hyperparameter combination corresponding to the optimal value of the fitness; If the preset conditions are not met, the step of updating the positions of the discoverer, joiner, and scout is performed; Satisfy the preset conditions and determine the optimal hyperparameter combination.

[0015] Optionally, in the step of extracting the electric shock current from the residual current signal using a pre-built electric shock current extraction model, the training method of the electric shock current extraction model is as follows: Collect data sets for training electric shock current extraction models; Construct the initial model of the gradient boosting decision tree algorithm XGBoost: define the loss function, define the penalty function, and initialize the hyperparameter combination of the initial model of the gradient boosting decision tree algorithm XGBoost; According to the data set, the initial model of the gradient boosting decision tree algorithm XGBoost is trained; during the training process, the grid search and cross-validation GSCV optimization algorithm are used to optimize the hyperparameter combination of the initial model of the gradient boosting decision tree algorithm XGBoost; After the training is completed, the optimal model of the gradient boosting decision tree algorithm XGBoost is obtained as the electric shock current extraction model.

[0016] Optionally, the specific steps for optimizing the initial model of the gradient boosting decision tree algorithm XGBoost using the grid search and cross-validation GSCV optimization algorithm are as follows: Determine the search range for each hyperparameter combination; Based on the search range, traverse all hyperparameter combinations and build a gradient boosting decision tree algorithm XGBoost model for each hyperparameter combination; Cross-validation was used to evaluate the performance of each gradient boosting decision tree algorithm XGBoost model, and the optimal root mean square error RMSE was recorded; The gradient boosting decision tree algorithm XGBoost model corresponding to the optimal root mean square error RMSE is used as the electric shock current extraction model.

[0017] Optionally, in the step of collecting the residual current signal on the user side, the residual current is a nonlinear signal, so when establishing the SVM recognition and classification model, it is necessary to introduce a nonlinear kernel function , maps fault data from low-dimensional space to high-dimensional feature space, constructs Lagrangian function to obtain the optimal classification hyperplane, has the advantages of wide convergence domain, fast convergence speed, and relatively simple parameter adjustment. The classification performance of the RBF kernel function mainly depends on the kernel function width and the penalty factor ; Hyperparameter combination ( , ) is optimal, the classification performance of the SVM model is also optimal. In order to obtain the optimal parameter combination of the SVM model, the SSA algorithm is used to and Optimize to improve the generalization and recognition accuracy of the model.

[0018] Optionally, the leakage fault safety protection criteria include human electric shock safety protection and photovoltaic abnormal leakage safety protection.

[0019] Optionally, the photovoltaic safety protection criteria include: obtaining the residual current signal, when the residual current signal is less than 30mA, the protection device does not alarm and does not act, and maintains a normal monitoring state; when the residual current signal reaches 30mA and is less than 60mA, the protection device sends an alarm signal but does not act; when the residual current signal reaches 60mA and is less than 100mA, the protection starts timing and sends an alarm signal; if the timing duration does not exceed 1min, the protection does not act; if the timing time reaches 1min, the protection trips instantly; when the residual current signal reaches 100mA, the protection acts immediately and sends an alarm signal; The criteria for electric shock safety protection include: when the electric shock current is less than 5mA, the protection device will not alarm and will not operate, maintaining a normal detection state; when the electric shock current reaches 5mA and is less than 30mA, the protection starts timing and sends out an alarm signal; if the timing duration does not exceed 1s, the protection will not operate; if the timing time reaches 1s, the protection will trip instantaneously; when the electric shock current reaches 30mA, the protection will operate immediately and send out an alarm signal.

[0020] In a second aspect, the present invention provides a leakage fault identification and electric shock protection system for a low-voltage distribution network, characterized in that it comprises: A collection module, used to collect residual current signals on the user side; A conversion module, for decomposing the residual current signal into IMF components based on VMD; The calculation module is used to calculate the total energy of the IMF components and use the energy values ​​of different frequency bands as features; An identification module inputs the features into a pre-built leakage fault type identification model, wherein the leakage fault type identification model identifies different leakage scenarios; The execution module is used to execute the protection criteria based on the leakage scene recognition result.

[0021] In a third aspect of the present invention, an electronic device is provided, comprising a processor and a memory, wherein the processor is used to execute a computer program stored in the memory to implement the electric shock protection method for the low-voltage distribution network.

[0022] According to a fourth aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores at least one instruction, and when the at least one instruction is executed by a processor, the method for protecting against electric shock in a low-voltage distribution network is implemented.

[0023] Compared with the prior art, the beneficial effects of the present invention are as follows: The electric shock current component can be accurately extracted from the complex residual current waveform to reflect the degree of the electric shock accident. After obtaining the electric shock current, different action measures are performed according to the effective value of the current, effectively preventing the RCD protection from false operation due to interference factors and refusal to operate in the electric shock state.

[0024] The RBF kernel function is selected, which has the advantages of wide convergence domain, fast convergence speed, and relatively simple parameter adjustment. Its classification performance mainly depends on the width of the kernel function. and the penalty factor , hyperparameter combination ( , ) is optimal, the classification performance of the SVM model is also optimal.

[0025] Using SSA algorithm and The optimization was performed to obtain the optimal parameter combination of the SVM model, which improved the generalization and recognition accuracy of the model.

[0026] The present invention implements the classification method through electronic devices and utilizes a processor to execute a computer program in a memory. The method can be efficiently run on modern electronic devices and is easy to promote and apply.

[0027] The computer-readable storage medium of the present invention stores instructions for implementing the classification method, so that the method can be easily adopted and executed by various computer systems, thereby improving the popularity and flexibility of the method. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The drawings constituting a part of the present application are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings: Figure 1 A flow chart of an embodiment of an electric shock protection method for a low-voltage distribution network of the present invention; Figure 2 The user-side residual current waveforms under three leakage scenarios of an electric shock protection method embodiment of a low-voltage distribution network of the present invention; Figure 3 The VMD decomposition results of the residual current signal on the user side under three scenarios of an electric shock protection method embodiment of a low-voltage distribution network of the present invention; Figure 4 This is a SSA-SVM modeling method flow of an electric shock protection method embodiment of a low-voltage distribution network of the present invention; Figure 5 This is a GSCV-XGBoost modeling method flow of an electric shock protection method embodiment of a low-voltage distribution network of the present invention; Figure 6 It is the overall design process of different leakage fault safety protection criteria of an electric shock protection method embodiment of a low-voltage distribution network of the present invention; Figure 7 An embodiment of an electric shock protection method for a low-voltage distribution network of the present invention is based on the action criterion and detailed implementation process of the integrated human-photovoltaic safety protection; Figure 8 Comparison of the recognition results of the SSA-SVM model in the present invention and the traditional KELM model for leakage type scenarios; Fig. 9 Comparison of the human body electric shock current extraction results of the GSCV-XGBoost model and the RF model in the present invention; Fig.10 A structural block diagram of a leakage fault identification and electric shock protection system for a low-voltage distribution network implemented by the present invention; Fig.11 The present invention is a structural block diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0029] The present invention will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that the embodiments and features in the embodiments of the present application can be combined with each other without conflict.

[0030] The following detailed description is an exemplary description, which is intended to provide further detailed description of the present invention. Unless otherwise specified, all technical terms used in the present invention have the same meaning as those generally understood by those skilled in the art to which the present application belongs. The terms used in the present invention are only for describing specific embodiments, and are not intended to limit exemplary embodiments according to the present invention.

[0031] It should be noted that the meanings of the abbreviations in the specification are: VDM: Variational Mode Decomposition, variational mode decomposition method IMF: Intrinsic Mode Function, ‌Intrinsic Mode Function SVM: Support Vector Machine SSA: Sparrow Search Algorithm, Sparrow Search Algorithm GSCV: Grid Search, Cross-Validation, Grid Search and Cross-Validation XGBoost: Extreme Gradient Boosting, gradient boosting decision tree algorithm KELM: Kernelized Extreme Learning Machine, based on kernelized extreme learning machine Example 1 like Figure 1 As shown, a method for protecting against electric shock in a low-voltage distribution network comprises the following specific steps: S1, collecting the residual current signal on the user side; S2, decomposing the residual current signal into IMF quantities using VMD; S3, calculating the total energy of the IMF component and normalizing it, and using the energy values ​​of different frequency bands of the normalized signal as features; S4, inputting the features into a pre-built leakage fault type recognition model to identify different leakage scenarios; When the identification result is a normal operating state, it switches to S1 to keep collecting residual current signals, and the protection does not operate; When the identification result is abnormal photovoltaic leakage, the process switches to S7 to execute photovoltaic safety protection criteria; When the identification result is an electric shock fault, the process switches to S6 to extract the electric shock current from the residual current; The leakage fault type identification model is constructed based on SSA and SVM.

[0032] S5. When the identification result is an electric shock fault, the electric shock current in the residual current signal is extracted using a pre-built electric shock current extraction model, and the electric shock safety protection criterion is executed based on the electric shock current; wherein the electric shock current extraction model is a GSCV-XGBoost regression analysis model.

[0033] As a preferred example, in S2, the VMD algorithm presets the number of components , Penalty Factor Two parameters, which set is 4, The optimal decomposition result is obtained when the value is 3000.

[0034] As a preferred example, in S3, the specific method for calculating the features is as follows: S1-1: Calculate the total energy of each IMF component: (1) In the formula, is the IMF component u k (t) Length; S1-2: Normalize the total energy of the IMF component: (2) S1-3: Construct a feature vector with the total energy of the IMF component as the element as the feature of the residual current signal: .

[0035] In step S4, the residual current is a nonlinear signal, so when establishing the SVM recognition and classification model, it is necessary to introduce a nonlinear kernel function , maps fault data from low-dimensional space to high-dimensional feature space, constructs Lagrangian function to obtain the optimal classification hyperplane, has the advantages of wide convergence domain, fast convergence speed, and relatively simple parameter adjustment. The classification performance of the RBF kernel function mainly depends on the kernel function width and the penalty factor ; Hyperparameter combination ( , ) is optimal, the classification performance of the SVM model is also optimal. In order to obtain the optimal parameter combination of the SVM model, the SSA algorithm is used to and Optimize to improve the generalization and recognition accuracy of the model.

[0036] like Figure 4 As shown, in step S4, the training method of the leakage fault type recognition model is as follows: S2-1: Obtain a training data set; wherein the training data set includes three categories of data, corresponding to different scenarios: data corresponding to normal operation, data corresponding to abnormal photovoltaic leakage, and data corresponding to electric shock failure; S2-2: Construct the initial SVM model, including constructing the Lagrangian function and the nonlinear kernel function , and setting hyperparameter combinations ( , )’s initial range; S2-3: Train the SVM initial model based on the training data set; during the training process, the SSA algorithm is used to adjust the hyperparameter combination ( , ) to optimize and obtain the optimal hyperparameter combination ( , ) to improve the generalization and recognition accuracy of the model; after the model converges, the leakage fault type recognition model is obtained.

[0037] like Figure 4 As shown, the SSA algorithm is used to optimize the hyperparameter combination ( , ) is optimized, the specific steps are: S3-1: Initialize the sparrow search algorithm SSA hyperparameter combination, calculate the fitness of the current hyperparameter combination, and calculate the accuracy of the cross-validation of the leakage fault type identification model under the current hyperparameter combination; S3-2: Update the positions of discoverers, joiners, and scouts; Specifically, the location is updated according to the following formula: Set sparrow population ,in , Indicated in In the dimensional space The location information of each sparrow in the d-dimensional space has a fitness of The entire population is divided into discoverers, joiners and scouts when foraging.

[0038] The function that defines the fitness is the inverse of the cross-validation accuracy of the model after each optimization, that is, .

[0039] To avoid the influence of some extreme values ​​on the algorithm results, the normalized fitness is calculated by the following formula:

[0040] The discoverer leads the entire population to continuously determine the location of food, accounting for 10% to 20% of the entire population. The location information is updated and iterated quickly. The location information of each iteration is as follows: (3) Where: Indicates the current iteration number; is the maximum number of iterations; Indicated in At the iteration, Only the discoverer The information value of the dimension; ; ; Indicates the warning value when danger occurs; is a random number that follows a normal distribution; For one The matrix whose elements are all 1. When it is judged that there is no danger, you can continue to forage. When danger appears, the discoverer needs to lead the population away to avoid being preyed upon by natural enemies.

[0041] The remaining sparrows in the population are joiners, and their position information is updated with iterations as follows: (4) Where: Indicates the best position currently occupied; It indicates the worst position of the entire population at present; For one The matrix of .when , some sparrows have very little food reserves, and they will leave and find a new place to find food.

[0042] When danger appears, the scout that senses danger will take anti-predator action, and its iterative position information is as follows: (5) Where: The optimal position for the whole; is a parameter with a mean of 0 and a variance of 1; Indicates the direction in which the sparrow moves; For the Normalized fitness of sparrows; is the best and worst fitness of all sparrows at present; is a constant value to avoid the situation where there is no solution.

[0043] when At this time, the sparrows are at the edge of the population and are very easy to be preyed upon by natural enemies; when When the scouts find their natural enemies, the sparrow population will adjust their positions to avoid risks.

[0044] S3-3: recalculate the fitness and determine the optimal value of the fitness; determine whether the accuracy of the cross-validation of the leakage fault type identification model meets the preset conditions under the hyperparameter combination corresponding to the optimal value of the fitness; S3-4: If the preset conditions are not met, the process goes to S3-2 to update the positions of the discoverer, joiner, and scout; S3-5: Satisfy the preset conditions and determine the optimal hyperparameter combination.

[0045] In step S6, Figure 5 As shown in the figure, the training method of the GSCV-XGBoost regression analysis model is as follows: S4-1: Obtain the training data set for GSCV-XGBoost regression analysis model training; S4-2: Build the initial XGBoost model; including defining the loss function, defining the penalty function, and initializing the XGBoost hyperparameter combination; S4-3: Perform model training based on the training data set. During the training process, use the GSCV optimization algorithm to optimize the XGBoost hyperparameter combination. When the model converges, obtain the XGBoost optimal model as the GSCV-XGBoost regression analysis model.

[0046] Optionally, after the model training is completed, the model is tested and tuned based on the test data set.

[0047] As a preferred example, the GSCV optimization algorithm is used to optimize the XGBoost hyperparameter combination. The specific steps are: S5-1: Define the search space for each parameter; the parameters include: the maximum depth of the tree, the learning rate, the minimum weight of the leaf node, and the penalty coefficient; S5-2: Set the search space for each hyperparameter combination. The grid search will traverse all hyperparameter combinations and compare the optimal RMSE. S5-3: Select a cross-validation strategy to evaluate the performance of each hyperparameter combination. The present invention adopts 10-fold cross-validation; S5-4: Use GSCV to optimize hyperparameter combinations. GSCV trains each combination in the hyperparameter combination space and evaluates the performance of the model through 10-fold cross validation. The core is to traverse all possible hyperparameter combinations and select the optimal RMSE based on the cross validation results. First, set the hyperparameter grid, and each point of the hyperparameter grid corresponds to a hyperparameter combination. Next, define the XGBoost model. Create a grid search cross validation object, pass in the XGBoost model, hyperparameter grid, and 10-fold cross validation parameters. Run the grid search and find the best hyperparameter combination.

[0048] S5-5: Determine that the result is not the optimal RMSE, and go to S5-2 to traverse the grid and compare the optimal RMSE; S5-6: Determine if the result is the optimal RMSE, retrain the XGBoost model, obtain the optimal XGBoost model, and evaluate its performance on the test set.

[0049] As a preferred example, Figure 6 As shown, in S6, the leakage fault safety protection criteria include human electric shock safety protection and photovoltaic abnormal leakage safety protection.

[0050] As a preferred example, Figure 7As shown in the figure, the specific steps of the human body electric shock safety protection implementation strategy are as follows: The photovoltaic safety protection criteria include: obtaining the residual current signal. When the residual current signal is less than 30mA, the protection device does not alarm and does not act, and maintains a normal monitoring state; when the residual current signal reaches 30mA and is less than 60mA, the protection device issues an alarm signal but does not act; when the residual current signal reaches 60mA and is less than 100mA, the protection starts timing and issues an alarm signal; if the timing duration does not exceed 1min, the protection does not act; if the timing time reaches 1min, the protection trips instantly; when the residual current signal reaches 100mA, the protection acts immediately and issues an alarm signal; The criteria for electric shock safety protection include: when the electric shock current is less than 5mA, the protection device will not alarm and will not operate, maintaining a normal detection state; when the electric shock current reaches 5mA and is less than 30mA, the protection starts timing and sends out an alarm signal; if the timing duration does not exceed 1s, the protection will not operate; if the timing time reaches 1s, the protection will trip instantaneously; when the electric shock current reaches 30mA, the protection will operate immediately and send out an alarm signal.

[0051] Figure 2 1 and 2 are the user-side residual current waveforms under three leakage scenarios in the embodiments of the present invention, where (a) is the residual current waveform in normal operating state, (b) is the residual current waveform in abnormal photovoltaic leakage, and (c) is the residual current waveform in electric shock fault.

[0052] Under normal operating conditions, the residual current signal amplitude is small and concentrated in the medium and low frequency bands of 50Hz~850Hz, with relatively few high-frequency components; when an electric shock fault occurs, the signal components are mainly concentrated in the low frequency band of 50Hz, accompanied by a small amount of medium and high frequency components; when a photovoltaic abnormal leakage fault occurs, the signal is mainly distributed in the medium and high frequency bands between 150Hz~5000Hz. The increase in photovoltaic leakage current causes the residual current amplitude and effective value to increase significantly. The existing threshold-action RCD cannot distinguish between electric shock accidents and photovoltaic abnormal leakage.

[0053] Figure 3 These are the VMD decomposition results of the residual current signal on the user side in the three scenarios in the embodiments of the present invention. The VMD decomposition decomposes the three residual current signals into signals in four frequency bands: IMF1, IMF2, IMF3, and IMF4. There are obvious differences between the different signals, and there is no problem of modal aliasing. There is also no false component in the high frequency band. The energy value characteristics of each frequency band are calculated and normalized as a data set to describe different fault types.

[0054] Electric shock fault detection example The recognition results of the SSA-SVM model of the present invention and the traditional KELM model for leakage type scenes are compared: Figure 8 The recognition results of the SSA-SVM model and the KELM model for leakage type scenes are compared. The coordinates 1, 2, and 3 represent the three states of normal system operation, abnormal photovoltaic leakage, and human electric shock, respectively, with a total of 117 test sets. From the recognition results, it can be seen that both recognition models correctly recognize the second state (abnormal photovoltaic leakage), which is attributed to the significant difference between the sample characteristics under the abnormal photovoltaic leakage state and the other two types of sample characteristics; the recognition accuracy of the SSA-SVM model reaches 100%, and the errors of the KELM model are concentrated on the human electric shock and normal operation states. Specifically, the KELM model misclassifies 5 samples in normal operation as human electric shock faults.

[0055] The results of extracting human electric shock current by the GSCV-XGBoost model of the present invention and the RF model are compared: Fig. 9 The comparison of the human body electric shock current extraction results of the GSCV-XGBoost model and the RF model is shown in the figure. Fig.11 It can be concluded that the electric shock current extracted based on the RF model has a poor fitting effect near the current peak, and the error range between the extracted value and the actual value is -2.21mA~2.09mA, while the human body electric shock current extracted based on the XGBoost model has an error range between -0.35mA~0.42mA. The XGBoost model has higher stability than the RF model in the process of extracting electric shock current.

[0056] Example 2 like Fig.10 As shown, based on the same inventive concept as the above embodiment, the present invention also provides a leakage fault identification and electric shock protection system for a low-voltage distribution network, comprising: A collection module, used to collect residual current signals on the user side; A conversion module, for decomposing the residual current signal into IMF components based on VMD; The calculation module is used to calculate the total energy of the IMF components and use the energy values ​​of different frequency bands as features; An identification module inputs the features into a pre-built leakage fault type identification model, wherein the leakage fault type identification model identifies different leakage scenarios; The execution module is used to execute the protection criteria based on the leakage scene recognition result.

[0057] Example 3 like Fig.11 As shown, the present invention also provides an electronic device 100 for implementing a method for identifying leakage faults and protecting against electric shock in a low-voltage distribution network; The electronic device 100 includes a memory 101 , at least one processor 102 , a computer program 103 stored in the memory 101 and executable on the at least one processor 102 , and at least one communication bus 104 .

[0058] The memory 101 can be used to store a computer program 103. The processor 102 implements the steps of a method for protecting against electric shock in a low-voltage distribution network in Example 1 by running or executing the computer program stored in the memory 101 and calling data stored in the memory 101.

[0059] The memory 101 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area may store data created according to the use of the electronic device 100 (such as audio data), etc. In addition, the memory 101 may include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (SmartMediaCard, SMC), a secure digital (SecureDigital, SD) card, a flash card (FlashCard), at least one disk storage device, a flash memory device, or other non-volatile solid-state storage devices.

[0060] At least one processor 102 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The processor 102 may be a microprocessor or any conventional processor, etc. The processor 102 is the control center of the electronic device 100, and uses various interfaces and lines to connect various parts of the entire electronic device 100.

[0061] Example 4 If the module / unit integrated in the electronic device 100 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 present invention implements all or part of the process in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. Computer-readable media may include: any entity or device that can carry computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory and read-only memory (ROM, Read-Only Memory).

[0062] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0063] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0064] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0065] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0066] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0067] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for protecting against electric shock in a low voltage distribution network, characterized in that: The specific steps include: Collect residual current signal on the user side; Decomposing the residual current signal into IMF components based on VMD; Calculate the total energy of the IMF components and use the total energy of the IMF components as a feature; Inputting the features into a pre-trained leakage fault type recognition model, the leakage fault type recognition model recognizes different leakage scenarios to obtain leakage scenario recognition results; Based on the leakage scenario identification results, the protection criteria are executed.

2. The electric shock protection method for a low-voltage distribution network according to claim 1, characterized in that: Based on the leakage scene identification results, protection criteria are executed, including: When the recognition result is a normal operating state, the residual current signal continues to be collected and the protection does not operate; when the recognition result is a photovoltaic abnormal leakage, the photovoltaic safety protection criterion is executed; when the recognition result is an electric shock fault, the pre-built electric shock current extraction model is used to extract the electric shock current in the residual current signal, and the electric shock safety protection criterion is executed based on the electric shock current.

3. The electric shock protection method for a low voltage distribution network according to claim 1, characterized in that: In the step of inputting the features into the leakage fault type recognition model, the training method of the leakage fault type recognition model is as follows: Obtain training data set: The data set contains three categories of data: data in normal operation, data when photovoltaic power leakage occurs, and data when electric shock occurs, corresponding to different scenarios; Construct the initial model of support vector machine SVM: construct Lagrangian function and nonlinear kernel function , set the hyperparameter combination ( , )’s initial range; Train the support vector machine SVM initial model according to the training data set; During the training process, the sparrow search algorithm SSA is used to optimize the hyperparameter combination ( , ) for optimization; After the training is completed, the optimal support vector machine (SVM) model is obtained as the optimal leakage fault type identification model.

4. The electric shock protection method for a low voltage distribution network as claimed in claim 3, characterized in that: The SSA algorithm is used to optimize the hyperparameter combination ( , ) is optimized, the specific steps are: Initialize the sparrow search algorithm SSA hyperparameter combination, calculate the fitness of the current hyperparameter combination, and calculate the accuracy of the cross-validation of the leakage fault type identification model under the current hyperparameter combination; Updates the positions of discoverers, joiners, and scouts; Recalculate the fitness and determine the optimal value of the fitness; determine whether the accuracy of the cross-validation of the leakage fault type identification model meets the preset conditions under the hyperparameter combination corresponding to the optimal value of the fitness; If the preset conditions are not met, the step of updating the positions of the discoverer, joiner, and scout is performed; Satisfy the preset conditions and determine the optimal hyperparameter combination.

5. The electric shock protection method for a low voltage distribution network as claimed in claim 2, characterized in that: In extracting the electric shock current from the residual current signal using a pre-built electric shock current extraction model, the training method of the electric shock current extraction model is as follows: Collect data sets for training electric shock current extraction models; Construct the initial model of the gradient boosting decision tree algorithm XGBoost: define the loss function, define the penalty function, and initialize the hyperparameter combination of the initial model of the gradient boosting decision tree algorithm XGBoost; According to the data set, loss function and penalty function, the initial model of the gradient boosting decision tree algorithm XGBoost is trained; during the training process, the grid search and cross-validation GSCV optimization algorithm are used to optimize the hyperparameter combination of the initial model of the gradient boosting decision tree algorithm XGBoost; After the training is completed, the optimal model of the gradient boosting decision tree algorithm XGBoost is obtained as the electric shock current extraction model.

6. The electric shock protection method for a low voltage distribution network as claimed in claim 5, characterized in that: The specific steps of optimizing the initial model of the gradient boosting decision tree algorithm XGBoost using the grid search and cross-validation GSCV optimization algorithm are as follows: Determine the search range for each hyperparameter combination; Based on the search range, traverse all hyperparameter combinations and build a gradient boosting decision tree algorithm XGBoost model for each hyperparameter combination; Cross-validation was used to evaluate the performance of each gradient boosting decision tree algorithm XGBoost model, and the optimal root mean square error RMSE was recorded; The gradient boosting decision tree algorithm XGBoost model corresponding to the optimal root mean square error RMSE is used as the electric shock current extraction model.

7. The electric shock protection method for a low voltage distribution network as claimed in claim 2, characterized in that: The photovoltaic safety protection criteria include: obtaining the residual current signal. When the residual current signal is less than 30mA, the protection device does not alarm and does not act, and maintains a normal monitoring state; when the residual current signal reaches 30mA and is less than 60mA, the protection device issues an alarm signal but does not act; when the residual current signal reaches 60mA and is less than 100mA, the protection starts timing and issues an alarm signal; if the timing duration does not exceed 1min, the protection does not act; if the timing time reaches 1min, the protection trips instantly; when the residual current signal reaches 100mA, the protection acts immediately and issues an alarm signal; The criteria for electric shock safety protection include: when the electric shock current is less than 5mA, the protection device will not alarm and will not operate, maintaining a normal detection state; when the electric shock current reaches 5mA and is less than 30mA, the protection starts timing and sends out an alarm signal; if the timing duration does not exceed 1s, the protection will not operate; if the timing time reaches 1s, the protection will trip instantaneously; when the electric shock current reaches 30mA, the protection will operate immediately and send out an alarm signal.

8. An electric shock protection system for a low voltage distribution network, characterized in that: include: A collection module, used to collect residual current signals on the user side; A conversion module, for decomposing the residual current signal into IMF components based on VMD; A calculation module, used for calculating the total energy of the IMF components, and taking the total energy of the IMF components as a feature; An identification module, used to input the features into a pre-built leakage fault type identification model, wherein the leakage fault type identification model identifies different leakage scenarios; The execution module is used to execute the protection criteria based on the leakage scene recognition result.

9. An electronic device, characterized in that: It comprises a processor and a memory, wherein the processor is used to execute a computer program stored in the memory to implement the electric shock protection method for a low-voltage distribution network as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores at least one instruction, and when the at least one instruction is executed by the processor, the electric shock protection method for the low-voltage distribution network as described in any one of claims 1 to 7 is implemented.

Images