Power distribution network fault classification method and system

Abstract

The invention discloses a power distribution network fault classification method and system. The method comprises the following steps: establishing a target power distribution network fault model; obtaining first time domain signal data in the simulation result of the target power distribution network model; constructing a first fault data set according to the first time domain signal data; and training a first power grid fault classification model by using the first fault data set, and performing power distribution network fault classification according to the trained first power grid fault classification model. By simulating complex working conditions of a real power distribution network, the method can better adapt to various conditions in practical application. Besides, the variational mode decomposition algorithm and the ensemble learning algorithm are combined, so that various fault types can be accurately identified, and high calculation efficiency can be displayed when large-scale data are processed. Therefore, the power distribution network fault classification method and system have a wide application prospect in the field of intelligent power grids and new energy grid connection.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network fault classification, and in particular to a distribution network fault classification method and system. Background Art

[0002] The safe and reliable operation of the distribution network directly affects the power supply of end customers. The distribution network faces various threats and damages in actual operation, such as lightning and typhoons, which may directly affect the distribution network. The aging, improper design and defects of equipment in long-term operation will cause the risk of short circuits and increase the probability of distribution network failures. After a failure occurs, industrial production will stagnate, commercial activities will be interrupted, and huge economic losses will be caused. With the increase of urban population and the development of technology, users have an increasingly strong demand for stable power supply. The importance and complexity of the distribution network are growing, and the speed and accuracy of its fault classification are of great significance to the safe and stable operation of the power system, reducing economic losses and maintaining public safety.

[0003] The fault classification methods of distribution networks can basically be divided into classification methods based on steady-state quantities and classification methods based on transient quantities. The method based on steady-state quantities mainly identifies the fault characteristics of the power frequency signal to diagnose the fault type. The method based on transient quantities mainly relies on artificial selection of fault signal quantities and transient feature extraction, setting thresholds and identifying transient features to determine the fault type. These methods have performed well in traditional distribution networks. However, when multiple power sources are connected to the grid, a large number of nonlinear features will appear in the system, and the traditional distribution network artificial feature identification methods are greatly affected. Therefore, the traditional distribution network classification method has been difficult to meet the development needs of the new distribution network, and it is urgent to develop a fault classification algorithm suitable for distributed new energy grid connection. Summary of the invention

[0004] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0005] In view of the above existing problems, the present invention is proposed.

[0006] Therefore, the present invention provides a distribution network fault classification method and system, which can solve the problems mentioned in the background technology.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] In a first aspect, the present invention provides a distribution network fault classification method, comprising:

[0009] Establish target distribution network fault model;

[0010] Obtaining first time domain signal data in the simulation result of the target distribution network model;

[0011] Constructing a first fault data set according to the first time domain signal data;

[0012] The first fault data set is used to train a first power grid fault classification model, and distribution network fault classification is performed according to the trained first power grid fault classification model.

[0013] As a preferred solution of the distribution network fault classification method described in the present invention, wherein: the establishment of the target distribution network fault model includes:

[0014] The target distribution network model includes at least several different types of short-circuit faults;

[0015] The several different types of short circuit faults are different types of short circuit faults generated under several different working conditions.

[0016] As a preferred solution of the distribution network fault classification method of the present invention, wherein: the obtaining of the first time domain signal data in the simulation result of the target distribution network model includes:

[0017] Acquire first time domain signal data in the simulation result of the target distribution network model, and perform first preprocessing on the first time domain signal data;

[0018] The first preprocessing includes at least a first fault labeling.

[0019] As a preferred solution of the distribution network fault classification method described in the present invention, wherein: the several different working conditions at least include a capacitor switching working condition and a high-resistance grounding working condition;

[0020] The different types of short circuit faults include at least a single-phase short circuit to ground fault, a two-phase short circuit fault, a two-phase short circuit to ground fault and a three-phase short circuit fault.

[0021] As a preferred solution of the distribution network fault classification method of the present invention, wherein: constructing a first fault data set according to the first time domain signal data includes:

[0022] Acquire three-phase current characteristics of first time domain signal data;

[0023] Acquire a first characteristic matrix according to the three-phase current characteristics;

[0024] And obtain a first fault data set according to the first characteristic matrix.

[0025] As a preferred solution of the distribution network fault classification method of the present invention, wherein: constructing a first fault data set according to the first time domain signal data includes:

[0026] A variational mode decomposition algorithm is used to solve the three-phase current characteristics of the first time domain signal data and solve the IMF component;

[0027] Calculate the kurtosis corresponding to the IMF component and reconstruct the IMF component according to the kurtosis criterion;

[0028] Construct the first characteristic matrix according to the reconstructed IMF components;

[0029] A first fault data set is acquired according to the first characteristic matrix.

[0030] As a preferred solution of the distribution network fault classification method of the present invention, wherein: using the first fault data set to train the first power grid fault classification model includes:

[0031] Using the first fault data set as a training sample set for a first power grid fault classification model;

[0032] The output of the first power grid fault classification model is the distribution network fault classification probability, and the input of the first power grid fault classification model is the real-time three-phase current characteristics of the first time domain signal data of the target distribution network.

[0033] In a second aspect, the present invention provides a distribution network fault classification system, comprising:

[0034] A model building module, used to build a target distribution network fault model;

[0035] A data acquisition module, used to acquire first time domain signal data in the simulation result of the target distribution network model;

[0036] A data set acquisition module, used to construct a first fault data set according to the first time domain signal data;

[0037] A classification module is used to train a first power grid fault classification model using the first fault data set, and classify distribution network faults according to the trained first power grid fault classification model.

[0038] In a third aspect, the present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above-mentioned method when executing the computer program.

[0039] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the method described above when executed by a processor.

[0040] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention proposes a distribution network fault classification method and system, establishes a target distribution network fault model; obtains the first time domain signal data in the simulation result of the target distribution network model; constructs a first fault data set according to the first time domain signal data; uses the first fault data set to train the first power grid fault classification model, and classifies the distribution network fault according to the first power grid fault classification model after training. A distribution network model including complex working conditions such as capacitor switching and high-resistance grounding faults is built, and the nonlinear characteristics introduced to the distribution network system by multi-power grid connection are considered; the current signal characteristics are extracted by variational mode decomposition, which can effectively extract the time-frequency characteristics of the signal; the Adaboost combined with KNN integrated learning algorithm of the distribution network fault classification method has a good performance in the aforementioned time-frequency feature matrix, and can achieve 99.7% fault classification accuracy on the constructed fault data set; the distribution network fault classification system of the present invention can effectively improve the accuracy and efficiency of fault detection. By simulating the complex working conditions of the real distribution network, the system can better adapt to various situations in practical applications. In addition, the combination of the variational mode decomposition algorithm and the ensemble learning algorithm adopted by the system can not only accurately identify various fault types, but also show high computational efficiency when processing large-scale data. Therefore, the distribution network fault classification method and system of the present invention have broad application prospects in the fields of smart grid and new energy grid connection. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0042] Figure 1 A method flow chart of a distribution network fault classification method and system provided by an embodiment of the present invention;

[0043] Figure 2 A detailed flow chart of a distribution network fault classification algorithm of a distribution network fault classification method and system provided by one embodiment of the present invention;

[0044] Figure 3 A distribution network fault classification method and an intrinsic mode function diagram obtained by variational mode decomposition of the system provided by an embodiment of the present invention;

[0045] Figure 4 A flow chart of an integrated learning algorithm of a distribution network fault classification method and system provided by an embodiment of the present invention;

[0046] Figure 5An internal structural diagram of a computer device of a distribution network fault classification method and system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0047] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0048] Example 1

[0049] Reference Figure 1-Figure 5 , which is the first embodiment of the present invention, and provides a distribution network fault classification method and system, including:

[0050] In the existing related technologies, there are some deficiencies in fault detection and classification, such as low accuracy of fault detection under complex working conditions and low computational efficiency when processing large-scale data.

[0051] The present application provides a method that can effectively solve the above-mentioned problems. Next, multiple embodiments will be combined to explain in detail how to implement the distribution network fault classification method;

[0052] Figure 1 A method flow chart of a distribution network fault classification method and system is shown, including:

[0053] S101, establishing a target distribution network fault model;

[0054] In an embodiment of the present application, establishing a target distribution network fault model includes:

[0055] The target distribution network model includes at least several different types of short-circuit faults;

[0056] Several different types of short-circuit faults are different types of short-circuit faults generated under several different working conditions.

[0057] In an optional embodiment, different types of short-circuit faults may be single-phase short-circuit to ground faults, two-phase short-circuit faults, two-phase short-circuit to ground faults, and three-phase short-circuit faults, and may also include arc short-circuit faults: This type of fault usually occurs in the air gap or the surface of the insulating material due to breakdown caused by overvoltage, forming an arc discharge, which may manifest as an unstable state and be accompanied by temperature rise and other physical phenomena. Intermittent faults: This type of fault is characterized by periodic or non-periodic appearance and disappearance within a certain period of time, which may be caused by poor contact, loose connections, or other temporary electrical problems. Transient faults: This type of fault occurs and recovers in a short period of time, such as branches being burned immediately after contacting a live wire, or short circuits caused by small animals. In addition, the target distribution network model may also include non-short-circuit faults, such as capacitor switching faults and high-resistance grounding faults.

[0058] In an embodiment of the present application, several different operating conditions include at least capacitor switching conditions and high-resistance grounding conditions; different types of short-circuit faults include at least single-phase short-circuit grounding fault, two-phase short-circuit fault, two-phase short-circuit grounding fault and three-phase short-circuit fault.

[0059] In an optional embodiment, the several different working conditions may include other working conditions in addition to the capacitor switching working condition and the high-resistance grounding working condition, such as transformer failure, line disconnection failure, etc. The establishment of the target distribution network model is intended to simulate various fault conditions that may occur in the real world, so as to facilitate the subsequent collection of fault data and the training of the classification model.

[0060] In the embodiment of the present application, single-phase short-circuit grounding fault, two-phase short-circuit fault, two-phase short-circuit grounding fault and three-phase short-circuit fault under capacitor switching conditions and high-resistance grounding conditions are selected to establish a target distribution network fault model. Specifically, PSCAD / EMTDC can be used to build a distribution network model. Capacitor switching and high-resistance grounding faults are added to the distribution network model. Faults of different phases and different transition resistances are set.

[0061] It should be noted that the establishment of the target distribution network fault model can simulate the fault conditions of the real distribution network under various complex working conditions, thereby providing rich and diverse data for the training of the fault classification model. In this way, it can be ensured that the model has high adaptability and accuracy when facing actual power grid faults. In addition, the establishment of the target distribution network model also takes into account the nonlinear characteristics brought about by the connection of multiple power sources to the grid, which is often ignored in previous fault detection systems, resulting in low fault detection accuracy. By simulating these complex working conditions, the present invention enables the fault classification system to better identify and handle various fault types in practical applications, thereby improving the operational safety and reliability of the entire distribution network.

[0062] S102, obtaining first time domain signal data in the simulation result of the target distribution network model;

[0063] In an embodiment of the present application, obtaining first time domain signal data in the simulation result of the target distribution network model includes:

[0064] Acquire first time domain signal data in the simulation result of the target distribution network model, and perform first preprocessing on the first time domain signal data;

[0065] The first preprocessing includes at least a first fault labeling.

[0066] In an embodiment of the present application, the first preprocessing may be filtering the first time domain signal data to remove noise interference. The filtering process may use a low-pass filter, a high-pass filter or a band-pass filter, depending on the characteristics of the signal and the requirements of fault detection. For example, if the signal contains high-frequency noise, a low-pass filter can be used to remove these high-frequency components. In addition, methods such as sliding average filtering and median filtering can be used to further smooth the signal and reduce the possibility of misjudgment. Then, the data after the noise interference is removed is subjected to the first fault labeling, and then the data after the first fault labeling is normalized.

[0067] In an optional embodiment, the first fault labeling is to label the collected first time domain signal data with a corresponding fault type, for example, labels such as "fault type A" and "fault type B" may be used to facilitate subsequent data classification and analysis. Other labeling methods may also be used, for example, digital coding or color coding to distinguish different fault types.

[0068] In an embodiment of the present application, labels "AG", "BG", "CG", "AB", "AC", "BC", "AB-G", "AC-G", "BC-G", "ABC", and "ABC-G" are used to identify different types of short-circuit faults that may occur in a distribution network, wherein AG represents a single-phase short-circuit grounding fault of phase A (single-phase fault, the fault point is between phase A and ground), BG represents a single-phase short-circuit grounding fault of phase B (single-phase fault, the fault point is between phase B and ground), CG represents a single-phase short-circuit grounding fault of phase C (single-phase fault, the fault point is between phase C and ground), AB represents a two-phase short-circuit fault of phase A and phase B (two-phase fault, the fault point is between phase A and phase B), and AC represents a two-phase short-circuit fault of phase A and phase C (two-phase fault). Fault, the fault point is between phase A and phase C), BC represents a two-phase short-circuit fault between phase B and phase C (two-phase fault, the fault point is between phase B and phase C), AB-G represents a two-phase short-circuit grounding fault between phase A and phase B (two-phase grounding fault, the fault point is between phase A, phase B and ground), AC-G represents a two-phase short-circuit grounding fault between phase A and phase C (two-phase grounding fault, the fault point is between phase A, phase C and ground), BC-G represents a two-phase short-circuit grounding fault between phase B and phase C (two-phase grounding fault, the fault point is between phase B, phase C and ground), ABC represents a three-phase short-circuit fault (three-phase fault, the fault point is between phase A, phase B and phase C), ABC-G represents a three-phase short-circuit grounding fault (three-phase grounding fault, the fault point is between phase A, phase B, phase C and ground). The reason for the design of this application is to combine it with the subsequent three-phase current characteristics so as to achieve more accurate fault type identification in the fault classification model. In this way, it is ensured that the model has high adaptability and accuracy when facing real power grid faults.

[0069] It should be noted that obtaining the first time domain signal data in the simulation results of the target distribution network model can provide rich training data for the fault classification model, thereby improving the model's ability to identify complex fault conditions in the real world. By simulating various fault types, including but not limited to capacitor switching faults, high-resistance grounding faults, single-phase short-circuit grounding faults, two-phase short-circuit faults, two-phase short-circuit grounding faults, and three-phase short-circuit faults, it can be ensured that the fault classification system can accurately identify and handle various fault types in practical applications. In addition, the establishment of the target distribution network model also takes into account the nonlinear characteristics brought about by the connection of multiple power sources to the grid, which is often overlooked in previous fault detection systems, resulting in low fault detection accuracy. By simulating these complex working conditions, the present invention enables the fault classification system to better identify and handle various fault types in practical applications, thereby improving the operational safety and reliability of the entire distribution network.

[0070] S103, constructing a first fault data set according to the first time domain signal data;

[0071] In an embodiment of the present application, constructing a first fault data set according to the first time domain signal data includes:

[0072] Acquire three-phase current characteristics of first time domain signal data;

[0073] Acquire a first characteristic matrix according to the three-phase current characteristics;

[0074] And obtain a first fault data set according to the first characteristic matrix.

[0075] In an optional embodiment, a variational mode decomposition algorithm can be used to process the three-phase current features of the first time domain signal data into a first feature matrix, and other signal processing techniques, such as Fourier transform or wavelet transform, can also be used to extract features. The first feature matrices obtained by different operations are different, but the operation of obtaining the first fault data set according to the first feature matrix is ​​similar, and the steps of obtaining the first fault data set using a similar method should be within the scope of protection of this application.

[0076] In the embodiment of the present application, constructing the first fault data set according to the first time domain signal data further includes:

[0077] A variational mode decomposition algorithm is used to solve the three-phase current characteristics of the first time domain signal data and solve the IMF component;

[0078] Calculate the kurtosis corresponding to the IMF component and reconstruct the IMF component according to the kurtosis criterion;

[0079] Construct the first characteristic matrix according to the reconstructed IMF components;

[0080] A first fault data set is acquired according to the first characteristic matrix.

[0081] In the embodiment of the present application, variational mode decomposition can decompose the signal into the sum of a series of intrinsic mode functions, and different IMF components represent the vibration of the signal at different frequencies. The constraint model of variational mode decomposition is:

[0082]

[0083] Among them, K is the number of modes to be decomposed (a positive integer), u k is the kth intrinsic mode function after decomposition, ω k is the center frequency of each component, f is the original signal, δ(t) is the Dirac function, * is the convolution operator, For partial derivatives, j is the imaginary unit.

[0084] In an optional embodiment, the kurtosis corresponding to the IMF component is calculated. The kurtosis is a statistic that measures the sharpness of the signal distribution. For an impact signal or a signal containing a fault, the kurtosis value usually increases. The kurtosis criterion can be used as an indicator to identify the fault characteristics that may be contained in the signal, so it is very useful in fault diagnosis. The characteristics of each IMF component after VMD decomposition are evaluated according to the kurtosis criterion, and the components with higher kurtosis values ​​are selected. These components are likely to contain fault information. The IMF components ranked high according to the selected kurtosis values ​​are combined to construct the first characteristic matrix.

[0085]

[0086] Among them, x i is the signal value, is the signal mean, N is the sampling length, σ i is the standard deviation.

[0087] It should be noted that the constructed first characteristic matrix can effectively reflect the characteristics of the distribution network under different fault conditions. By analyzing the data in the first characteristic matrix, key information reflecting the fault characteristics can be extracted, such as the degree of distortion of the current waveform, changes in frequency components, etc. This information is crucial for subsequent fault classification because it can help the model distinguish different types of faults. For example, a single-phase short circuit fault usually causes obvious asymmetry in the current waveform, while a two-phase short circuit fault may manifest as an increase in specific frequency components of the current waveform. Through the extraction and analysis of these features, the fault classification model can more accurately identify the fault type, thereby providing strong support for the protection and restoration of the distribution network.

[0088] In an optional embodiment, a data set that satisfies the training of the power grid fault classification model can be directly extracted from the first feature matrix. The specific operation method can be to randomly select a part of the data as a training set and another part as a test set. It is also possible to use other methods for the first fault data set, for example, to ensure the diversity of the data set and the generalization ability of the model through a cross-validation method. Cross-validation is a statistical method that divides the data set into several small groups, takes turns using one group as a test set and the rest as a training set to evaluate the performance of the model on different data subsets. This method helps to reduce the risk of overfitting the model and improve the prediction accuracy of the model on unknown data.

[0089] In an embodiment of the present application, the three-phase current characteristics of the first time domain signal data are obtained. At this stage, the first time domain signal data is extracted from the simulation results of the target distribution network model, and special attention is paid to the three-phase current signal. A first characteristic matrix is ​​obtained according to the three-phase current characteristics, and the three-phase current characteristics are processed and converted into a characteristic matrix. The matrix contains a series of fault-related characteristics, such as the amplitude, phase, frequency and other information of the current. The three-phase current characteristics are processed by the variational mode decomposition algorithm, and the three-phase current characteristics are further analyzed using the variational mode decomposition (VMD) algorithm to obtain a set of intrinsic mode function (IMF) components. The VMD algorithm can separate components of different frequencies in the signal, which is helpful for the subsequent feature extraction. The kurtosis of the IMF component is calculated, and for each IMF component, its kurtosis is calculated. Kurtosis is a statistic used to measure the thickness of the tail of the distribution, which can help identify abnormal features in the signal. Reconstruct the IMF components according to the kurtosis criterion. Based on the calculation results of the kurtosis, select and reconstruct the IMF components according to certain criteria (such as IMF components with high kurtosis values ​​may contain more useful information) to retain the features that are most important for fault classification. Construct the first characteristic matrix based on the reconstructed IMF components, and integrate the reconstructed IMF components into a new characteristic matrix. This matrix should contain useful information that can distinguish different types of faults. Obtain the first fault data set based on the first characteristic matrix, and extract the data set that can be used to train the power grid fault classification model from the constructed first characteristic matrix. These data sets should contain clearly labeled fault types so that the training model can learn to identify different types of faults.

[0090] It should be noted that step S103 can significantly improve the accuracy and efficiency of fault detection. Through the variational mode decomposition algorithm, key fault features can be effectively extracted from complex distribution network signals, which may be difficult to identify in traditional methods. In addition, the application of the kurtosis criterion further optimizes the fault feature extraction process and ensures that the selected feature matrix contains the components that best represent the fault information. Ultimately, these carefully selected and reconstructed feature matrices provide high-quality training data for the fault classification model, so that the model can more accurately identify various fault types in practical applications, providing strong technical support for the safe and stable operation of the distribution network.

[0091] S104: Use the first fault data set to train a first power grid fault classification model, and perform distribution network fault classification according to the trained first power grid fault classification model.

[0092] In an embodiment of the present application, using the first fault data set to train the first power grid fault classification model includes:

[0093] Using the first fault data set as a training sample set for a first power grid fault classification model;

[0094] The output of the first power grid fault classification model is the distribution network fault classification probability, and the input of the first power grid fault classification model is the real-time three-phase current characteristics of the first time domain signal data of the target distribution network.

[0095] like Figure 2 As shown, 200 time domain signal is the starting point of the process, indicating that the input is the real-time current or voltage signal from the distribution system. 201 Use variational mode decomposition to extract three-phase current features and solve IMF components: Preprocess the time domain signal, use variational mode decomposition (VMD) technology to extract important features of three-phase current, and solve its intrinsic mode function (IMFs), which is the basis for understanding the signal structure. 202 Calculate the frequency corresponding to the IMF component and reconstruct the IMF component according to the frequency criterion: Analyze the obtained IMF components, determine their respective frequency components, and reorganize these components according to certain rules to ensure that the most relevant feature information is retained. 203 Construct a feature matrix based on the reconstructed IMF components: Convert the reorganized IMF components into a form suitable for further analysis - a feature matrix, which is convenient for subsequent machine learning operations. 204 High-performance distribution network fault classification algorithm: Apply advanced algorithms to conduct in-depth analysis of the feature matrix to identify and classify various faults in the distribution network. 205 Build an integrated learning algorithm combining Adaboost and KNN: Build a hybrid model that integrates Adaboost and K-nearest neighbor (KNN) algorithms, which is a powerful machine learning strategy that can effectively improve prediction accuracy. 206 Iteratively optimize model parameters: Continuously adjust and optimize the parameter configuration within the model to achieve the best classification effect. 207 Distribution network fault classification results: After a series of processing and analysis, the final distribution network fault classification results are output.

[0096] In an optional embodiment, as long as the output of the first power grid fault classification model is the distribution network fault classification probability, and the input of the first power grid fault classification model is the real-time three-phase current characteristics of the first time domain signal data of the target distribution network, then no matter what means are used to establish the model, it should be within the scope of protection of this application.

[0097] In an optional embodiment, a support vector machine (SVM) algorithm can be used to build a fault classification model. SVM is an effective classification method that distinguishes data points of different categories by finding an optimal hyperplane, thereby identifying the fault type. When training the SVM model, different kernel functions can be used to handle nonlinear problems, such as the radial basis function (RBF) kernel, which can map data to a higher dimensional space to better separate fault categories.

[0098] In an optional embodiment, cross-validation technology can also be used to evaluate the generalization ability of the model to ensure that the performance of the model on unknown data is stable and reliable. Through cross-validation, the model's dependence on a specific data set can be reduced and the robustness of the model can be improved. After the model training is completed, the test data set can be used to perform a final performance evaluation of the model to ensure the accuracy and reliability of the model in practical applications. Ultimately, through these steps, an efficient and accurate distribution network fault classification system can be constructed to provide a strong technical guarantee for the stable operation of the power grid.

[0099] In an optional embodiment, it is also possible to use an integrated learning algorithm of Adaboost combined with KNN according to the algorithm principle of Adaboost to train the aforementioned first fault data set of the distribution network.

[0100] In the embodiment of the present application, an ensemble learning algorithm combining Adaboost and KNN is selected to construct an ensemble learning algorithm combining Adaboost and KNN. First, the KNN algorithm is written in pytorch, and a callable interface is set in it. Then the Adaboost ensemble learning algorithm is written, and the weak classifier part of the ensemble learning algorithm calls the aforementioned KNN algorithm. The weights of the weak classifiers in the ensemble learning algorithm will be iteratively updated according to the classification results. The KNN algorithm has low time complexity, mature theory, and simple architecture. It is suitable as a weak classifier for preliminary training of data sets. Figure 4 As shown, data set: First, there is an original data set. Average weight: Assign an initial weight to each sample in the data set, usually uniformly distributed. Train weak classifiers: Use the weighted data set D1 to train the first weak classifier. After training, adjust the sample weights according to the performance of the weak classifier on the training set. The weights of incorrectly classified samples increase, and the weights of correctly classified samples decrease. Repeat step 3: Use the updated weights and data set D1 again to train the second weak classifier, and adjust the sample weights again. This process will be repeated many times, and each time a new weak classifier will be generated and the sample weights will be adjusted. Combine strong classifiers: Combine all trained weak classifiers to form a strong classifier. Usually, the output of each weak classifier is assigned different weights, which reflect the performance of the weak classifier on the training set. The final result: By combining the outputs of multiple weak classifiers, a stronger and more accurate strong classifier is obtained.

[0101] It should be noted that the principle of the KNN algorithm is relatively simple. It mainly uses a certain distance metric to find the k training samples closest to it in the training set, and then makes predictions based on the information of the k neighbors. The distance metric is

[0102]

[0103] When p = 1, the metric is Manhattan distance. When p = 2, the distance is called Euclidean distance.

[0104] Furthermore, according to the algorithm principle of Adaboost, an integrated learning algorithm combining Adaboost and KNN was built to train the above-mentioned distribution network fault data set, continuously optimize the model parameters, and improve the model accuracy.

[0105] In an optional embodiment, the model that completes parameter optimization is combined with variational modal decomposition, and the variational modal decomposition algorithm can be written in matlab, and the variational modal decomposition and the IMF component with high kurtosis value calculated according to the kurtosis criterion are combined to obtain a feature matrix. The feature matrix is ​​read with python and converted into floot32 format for storage. The feature matrix is ​​read and trained using the aforementioned Adaboost combined with KNN integrated learning algorithm, and the algorithm parameters are optimized by back propagation to achieve a high-accuracy distribution network fault classification algorithm.

[0106] In summary, this distribution network fault classification method builds a distribution network model with complex working conditions including capacitor switching and high-resistance grounding faults, and takes into account the nonlinear characteristics introduced by multi-power grid connection to the distribution network system; this distribution network fault classification method uses variational mode decomposition to extract current signal features, which can effectively extract the time-frequency features of the signal; the Adaboost combined with KNN integrated learning algorithm of this distribution network fault classification method has a good performance in the aforementioned time-frequency feature matrix, and can achieve a fault classification accuracy of 99.7% on the constructed fault data set.

[0107] This embodiment also provides a distribution network fault classification system, including:

[0108] A model building module, used to build a target distribution network fault model;

[0109] A data acquisition module, used to acquire first time domain signal data in the simulation result of the target distribution network model;

[0110] A data set acquisition module, used to construct a first fault data set according to the first time domain signal data;

[0111] The classification module is used to train a first power grid fault classification model using a first fault data set, and classify distribution network faults according to the trained first power grid fault classification model.

[0112] The above-mentioned unit modules may be embedded in or independent of a processor in a computer device in the form of hardware, or may be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to the above-mentioned modules.

[0113] This embodiment also provides a computer device, which may be a terminal, and its internal structure diagram may be as shown in FIG. Figure 5 As shown. The computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a distribution network fault classification method is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a key, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0114] This embodiment further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0115] Establish target distribution network fault model;

[0116] Obtaining first time domain signal data in the simulation result of the target distribution network model;

[0117] Constructing a first fault data set according to the first time domain signal data;

[0118] The first fault data set is used to train a first power grid fault classification model, and distribution network fault classification is performed according to the trained first power grid fault classification model.

[0119] Example 2

[0120] Reference Figure 2-Figure 4 , which is an embodiment of the present invention, provides a distribution network fault classification method and system. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through experiments.

[0121] Figure 2The detailed flow chart of the distribution network fault classification algorithm is shown. The distribution network model is built using PSCAD / EMTDC. Capacitor switching and high-resistance grounding faults are added to the distribution network model. Faults with different phases and different transition resistances are set. The power phases are set to (15°, 30°, 45°, 60°, 75°, 90°, 105°, 120°, 135°, 150°, 165°, 180°, 195°, 210°, 225°, 240°, 255°, 270°, 285°, 300°, 315°, 330°, 345°, 360°) during the fault. The fault resistance is set to (0.01Ω, 5Ω, 20Ω, 50Ω, 100Ω, 200Ω, 500Ω, 1000Ω), the model is simulated, and the sensor nodes are set to collect the fault current and store the data in the form of .out.

[0122] Use Matlab to read the fault data and label it with the corresponding fault type. Figure 3 As shown in the figure, the VMD algorithm is used to decompose the fault current and obtain the intrinsic mode functions with different center frequencies. The kurtosis value of the intrinsic mode function is calculated, and according to the kurtosis criterion, the IMF components with high kurtosis values ​​are combined to construct the fault feature matrix, which is stored in the form of .mat.

[0123] Use Python to read the fault feature matrix and build an Adaboost combined with KNN ensemble learning algorithm. The structure is as follows Figure 4 As shown in the figure, the fault data set is divided into a training set and a test set according to a ratio of 7:3 to train the built ensemble learning algorithm. The model can achieve an accuracy of more than 99.7% on both the training set and the test set. Then, the fault feature matrix is ​​randomly selected for verification, and it is found that the fault classification can be accurately achieved. The variational mode decomposition and the ensemble learning algorithm are combined to construct a complete distribution network fault classification algorithm. The time domain signal is calculated by the fault classification algorithm to obtain the fault classification result.

[0124] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

[0125] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of complete hardware embodiments, complete software embodiments, or embodiments in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code. The scheme in the embodiments of the present application can be implemented in various computer languages, for example, object-oriented programming language Java and literal scripting language JavaScript, etc.

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

[0127] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture 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.

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

[0129] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0130] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A distribution network fault classification method, characterized in that: include: Establish target distribution network fault model; Obtaining first time domain signal data in the simulation result of the target distribution network model; Constructing a first fault data set according to the first time domain signal data; The first fault data set is used to train a first power grid fault classification model, and distribution network fault classification is performed according to the trained first power grid fault classification model.

2. The distribution network fault classification method according to claim 1, characterized in that: The establishment of the target distribution network fault model comprises: The target distribution network model includes at least several different types of short-circuit faults; The several different types of short circuit faults are different types of short circuit faults generated under several different working conditions.

3. The distribution network fault classification method according to claim 2, characterized in that: The obtaining of the first time domain signal data in the simulation result of the target distribution network model comprises: Acquire first time domain signal data in the simulation result of the target distribution network model, and perform first preprocessing on the first time domain signal data; The first preprocessing includes at least a first fault labeling.

4. The distribution network fault classification method according to claim 3, characterized in that: The several different working conditions at least include a capacitor switching working condition and a high-resistance grounding working condition; The different types of short circuit faults include at least a single-phase short circuit to ground fault, a two-phase short circuit fault, a two-phase short circuit to ground fault and a three-phase short circuit fault.

5. The distribution network fault classification method according to claim 4, characterized in that: The constructing a first fault data set according to the first time domain signal data comprises: Acquire three-phase current characteristics of first time domain signal data; Acquire a first characteristic matrix according to the three-phase current characteristics; And obtain a first fault data set according to the first characteristic matrix.

6. The distribution network fault classification method according to claim 5, characterized in that: The constructing a first fault data set according to the first time domain signal data further comprises: A variational mode decomposition algorithm is used to solve the three-phase current characteristics of the first time domain signal data and solve the IMF component; Calculate the kurtosis corresponding to the IMF component and reconstruct the IMF component according to the kurtosis criterion; Construct the first characteristic matrix according to the reconstructed IMF components; A first fault data set is acquired according to the first characteristic matrix.

7. The distribution network fault classification method according to claim 6, characterized in that: The using the first fault data set to train a first power grid fault classification model comprises: Using the first fault data set as a training sample set for a first power grid fault classification model; The output of the first power grid fault classification model is the distribution network fault classification probability, and the input of the first power grid fault classification model is the real-time three-phase current characteristics of the first time domain signal data of the target distribution network.

8. A distribution network fault classification system, characterized in that: include: A model building module, used to build a target distribution network fault model; A data acquisition module, used to acquire first time domain signal data in the simulation result of the target distribution network model; A data set acquisition module, used to construct a first fault data set according to the first time domain signal data; A classification module is used to train a first power grid fault classification model using the first fault data set, and classify distribution network faults according to the trained first power grid fault classification model.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images