Power distribution network section positioning method based on edge collaboration and artificial intelligence
By adopting edge collaboration and artificial intelligence methods in the distribution network, using multi-node collaborative work and voting mechanisms for fault location, the problems of low accuracy and slow processing speed in the existing technology are solved, and a higher Accuracy and reliability, and shorten troubleshooting time.
Patent Information
- Application Number
- CN202411075520.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-07
- Publication Date
- 2024-11-19
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing distribution network fault positioning algorithm has low accuracy and slow processing speed, which cannot guarantee fastness and reliability, and cannot effectively realize rapid positioning and isolation of fault segments.
Using edge collaboration and artificial intelligence methods, edge nodes are partitioned and set up edge nodes on the distribution network, fault location is performed through multi-node collaborative work and voting mechanisms, and data preprocessing and noise reduction are used to use current transformer data and ICEEMDAN signal decomposition technology. Build a global graph and independently judge the fault segment, and finally determine the final result through the voting system.
It improves the accuracy and reliability of fault location, shortens fault processing and recovery time, and enhances the system's adaptability and generalization capabilities.
Smart Images

Figure CN118980886A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of distribution network fault diagnosis, and in particular to a distribution network section positioning method based on edge collaboration and artificial intelligence. Background Art
[0002] With the development of my country's distribution network, the network intelligence of distribution systems across the country has accelerated. In order to realize feeder automation management, and with the development of information technology, feeder automation management systems combined with information technology have also been introduced into distribution networks. However, with the improvement of people's living standards, the dependence on electricity in daily life is also increasing, and higher requirements are placed on the safety and reliability of distribution networks. One of the measures to ensure the safe and stable operation of distribution networks is to be able to quickly locate and isolate faults and quickly restore power supply to non-fault areas.
[0003] Scholars at home and abroad have proposed a variety of methods for locating sections of feeder automation distribution networks. Some methods use genetic algorithms for fault location, but the genetic algorithm itself requires a lot of calculations, which is inconsistent with the rapid positioning required by the power system. Some methods use matrix algorithms for fault location. The matrix algorithm has a simple principle and is suitable for simple distribution networks, but its robustness is poor. Some methods use swarm intelligence algorithms such as particle swarms, ant colonies, and chicken swarms. If you want to achieve global optimality, you need to distribute enough swarm individuals in the algorithm initialization stage, which will also make the search process extremely time-consuming. If the number of swarm individuals distributed during initialization is insufficient, it is easy to fall into local optimality.
[0004] Most of the above methods cannot guarantee the accuracy of fault section location, and the above methods transmit information to the main station for processing, which has low processing efficiency, high signal transmission pressure, and cannot guarantee speed. Therefore, it is necessary to study the fault location algorithm based on the above problems. Summary of the invention
[0005] The purpose of the present invention is to overcome the shortcomings of the fault section locating algorithm in the prior art, such as low accuracy, slow processing speed and difficult reliability guarantee, and to provide a distribution network section locating method based on edge collaboration and artificial intelligence.
[0006] The purpose of the present invention is achieved through the following technical solutions: A distribution network section positioning method based on edge collaboration and artificial intelligence includes the following steps: Step 1: partition the distribution network and set edge nodes; Step 2: When a fault occurs in the distribution network, each node in the partition collects the fault current and uploads it to the edge node of the corresponding partition; Step 3: The edge nodes of each partition aggregate the collected fault current information and share it with the edge nodes of other partitions; Step 4: After receiving the shared data, each edge node constructs a global recognition map and inputs it into its own fault location model. Step 5: Each edge node independently determines the fault section through its own fault location model and uploads the location result to the voting system. Step 6: The voting system receives the summarized positioning results and performs majority voting to obtain the final fault section; Step 7: The voting system uploads the final fault section location result to the central control system, and the central control system handles the fault.
[0007] Preferably, the step 1 is specifically as follows: The distribution network is partitioned by using a distribution network partitioning algorithm based on current transformer data. First, the current data of each node in the distribution network is collected and preprocessed. Then, the correlation matrix between each node is calculated, and the correlation degree of each node in the distribution network is calculated. Construct the Laplace matrix L, perform eigenvalue decomposition on the Laplace matrix L, obtain the eigenvectors corresponding to the first k smallest eigenvalues, form the eigenvector matrix U, use the K-means clustering algorithm to cluster the matrix U, divide the nodes into several sub-segments, and take the cluster center or the point closest to the cluster center as the edge node; According to the clustering results, the segment to which each node belongs is determined, and the distribution network partition is completed.
[0008] Preferably, the calculation of the association matrix between the nodes and the calculation of the association degree of each node in the distribution network are specifically as follows: The current correlation coefficient is used to represent the node correlation. Let the currents of node n and node m be I n (t) and I m (t), correlation coefficient R nm for: In the above formula, and are the average current values of node n and node m in time T respectively; According to the above correlation coefficient R nm , construct the correlation matrix R, the matrix R is an N×N matrix, where the element R nm Represents the association between node n and node m, where N is the number of nodes; The association graph G(V, E) can be obtained from the association matrix, where V is the node in the distribution network and E is the association edge between nodes. nm Exceeding a certain threshold R th When , there is an edge between node n and node m.
[0009] Preferably, the expression of the Laplace matrix L is: L=DA Where D is the degree matrix, that is, the diagonal matrix of the number of edges connecting the nodes; A is the adjacency matrix, that is, the matrix composed of the edges between nodes.
[0010] Preferably, in step 2, the collected fault current is further decomposed and denoised, specifically: ICEEMDAN is used to decompose the current signal. First, the original current signal x[n] is taken and Gaussian white noise is added to construct a total of I groups of noisy signals. Let w (i) is Gauss white noise with mean 0 and unit variance; x (i) =x+β0E1(w (i) ) In the formula, w (i) represents the i-th group of white noise added, where i = 1, 2, …, I; the operator E1(·) represents the first IMF component of the EMD decomposition of a signal, β0 represents the weight coefficient when the noise component is added, which represents the ratio of the signal-to-noise ratio of the added noise to the standard deviation of the noise component; The local operator M(·) is introduced to represent the local mean of the signal. The local mean is the part obtained by subtracting this IMF from the original signal. The calculation can get each x (i) The corresponding first-order IMF component Summing the above components, we can get the first-order IMF component D1 of the zero-sequence current signal x, The first-order residual r1 can be obtained. r1=x-D1 The second-order residual r2 is estimated as r1+β1E2(w (i) ), The second-order IMF component D2 is obtained as Repeat the above steps to get the k-th order residual r k (k=3,4,…,KK is the total decomposition order) The Kth order IMF component D k for D k =r k-1 -r k Repeat the above steps to obtain the IMF components decomposed by ICEEMDAN.
[0011] Preferably, the distribution network section positioning method based on edge collaboration and artificial intelligence also includes step 8, feeding back the final fault section positioning result to each edge node, and each edge node uses the final fault section positioning result as real data to update the fault positioning model.
[0012] The beneficial effects of the present invention are: (1) improving accuracy: through multi-node collaborative work and voting mechanism, single-node misjudgment can be effectively avoided and the accuracy of fault location can be improved.
[0013] (2) Enhanced reliability: Each node independently detects and judges, and the final result is determined through a voting mechanism, which enhances the reliability and fault tolerance of the system.
[0014] (3) Achieve rapid response: Through the collaborative work of multiple nodes, faults can be quickly detected and located, shortening fault handling and recovery time.
[0015] (4) Adaptive optimization: Through data feedback and voting result analysis, the fault detection and location algorithm is continuously optimized and improved to enhance the system's adaptability and generalization capabilities. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] Figure 1 is a flow chart of the present invention; Figure 2 It is a schematic diagram of the IEEE33 node distribution network model; Figure 3 It is a schematic diagram of IEEE33 node distribution network partition; Figure 4 GoogLeNet network structure diagram. DETAILED DESCRIPTION
[0017] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this application will be more comprehensive and complete and fully convey the concept of the example embodiments to those skilled in the art.
[0018] In addition, described feature, structure or characteristic can be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present application. However, those skilled in the art will appreciate that the technical scheme of the present application can be put into practice without one or more of the specific details, or other methods, components, devices, steps, etc. can be adopted. In other cases, known methods, devices, realizations or operations are not shown or described in detail to avoid blurring the various aspects of the application.
[0019] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities may be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.
[0020] The flowcharts shown in the accompanying drawings are only exemplary and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps can be decomposed, and some operations / steps can be combined or partially combined, so the actual execution order may change according to actual conditions.
[0021] Example: A distribution network section positioning method based on edge collaboration and artificial intelligence, such as Figure 1 As shown, the following steps are included: Step 1: partition the distribution network and set edge nodes; Step 2: When a fault occurs in the distribution network, each node in the partition collects the fault current and uploads it to the edge node of the corresponding partition; Step 3: The edge nodes of each partition aggregate the collected fault current information and share it with the edge nodes of other partitions; Step 4: After receiving the shared data, each edge node constructs a global recognition map and inputs it into its own fault location model. Step 5: Each edge node independently determines the fault section through its own fault location model and uploads the location result to the voting system. Step 6: The voting system receives the summarized positioning results and performs majority voting to obtain the final fault section; Step 7: The voting system uploads the final fault section location result to the central control system, and the central control system handles the fault; Step 8: Feedback the final fault section location result to each edge node, and each edge node uses the final fault section location result as real data to update the fault location model.
[0022] The step 1 is specifically as follows: The distribution network is partitioned by using a distribution network partitioning algorithm based on current transformer data. First, the current data of each node in the distribution network is collected and preprocessed. Then, the correlation matrix between each node is calculated, and the correlation degree of each node in the distribution network is calculated. Construct the Laplace matrix L, perform eigenvalue decomposition on the Laplace matrix L, obtain the eigenvectors corresponding to the first k smallest eigenvalues, form the eigenvector matrix U, use the K-means clustering algorithm to cluster the matrix U, divide the nodes into several sub-segments, and take the cluster center or the point closest to the cluster center as the edge node; According to the clustering results, the segment to which each node belongs is determined, and the distribution network partition is completed.
[0023] The calculation of the correlation matrix between each node and the correlation degree of each node in the distribution network is specifically as follows: the correlation degree of the node is represented by the current correlation coefficient, and the currents of node n and node m are I n (t) and I m (t), correlation coefficient R nm for: In the above formula, and are the average current values of node n and node m in time T respectively; According to the above correlation coefficient R nm , construct the correlation matrix R, the matrix R is an N×N matrix, where the element R nm Represents the association between node n and node m, where N is the number of nodes; The association graph G(V, E) can be obtained from the association matrix, where V is the node in the distribution network and E is the association edge between nodes. nm Exceeding a certain threshold R th When , there is an edge between node n and node m.
[0024] The expression of the Laplace matrix L is: L=DA Where D is the degree matrix, that is, the diagonal matrix of the number of edges connecting the nodes; A is the adjacency matrix, that is, the matrix composed of the edges between nodes.
[0025] In the step 2, the collected fault current is also decomposed and denoised, specifically: ICEEMDAN is used to decompose the current signal. First, the original current signal x[n] is taken and Gaussian white noise is added to construct a total of I groups of noisy signals. Let w (i) is Gauss white noise with mean 0 and unit variance; x(i) =x+β0E1(w (i) ) In the formula, w (i) represents the i-th group of white noise added, where i = 1, 2, …, I; the operator E1(·) represents the first IMF component of the EMD decomposition of a signal, β0 represents the weight coefficient when the noise component is added, which represents the ratio of the signal-to-noise ratio of the added noise to the standard deviation of the noise component; The local operator M(·) is introduced to represent the local mean of the signal. The local mean is the part obtained by subtracting this IMF from the original signal. The calculation can get each x (i) The corresponding first-order IMF component Summing the above components, we can get the first-order IMF component D1 of the zero-sequence current signal x, The first-order residual r1 can be obtained. r1=x-D1 The second-order residual r2 is estimated as r1+β1E2(w (i) ), The second-order IMF component D2 is obtained as Repeat the above steps to get the k-th order residual r k (k=3,4,…,KK is the total decomposition order) The Kth order IMF component D k for D k =r k-1 -r k Repeat the above steps to obtain the IMF components decomposed by ICEEMDAN.
[0026] like Figure 2 As shown in the figure, this scheme is applied to the IEEE 33-node distribution network model for verification, and the partitioning algorithm mentioned above is used for partitioning. The partitioning results are as follows: Figure 3 As shown, node 3, node 29, and node 12 are selected as edge nodes, and a fault location model is configured on these edge nodes.
[0027] The fault location model of this solution adopts the GoogLeNet neural network model. Figure 4This is the network structure diagram of GoogLeNet, which consists of multiple convolutional layers, Inception modules and pooling layers. The Inception module consists of 1×1 convolution + 3×3 convolution + 5×5 convolution + 3×3 maximum pooling.
[0028] By adopting the above scheme, the following advantages and beneficial effects can be achieved: (1) Improved accuracy: Through multi-node collaboration and voting mechanisms, single-node misjudgment can be effectively avoided and the accuracy of fault location can be improved.
[0029] (2) Enhanced reliability: Each node independently detects and judges, and the final result is determined through a voting mechanism, which enhances the reliability and fault tolerance of the system.
[0030] (3) Achieve rapid response: Through the collaborative work of multiple nodes, faults can be quickly detected and located, shortening fault handling and recovery time.
[0031] (4) Adaptive optimization: Through data feedback and voting result analysis, the fault detection and location algorithm is continuously optimized and improved to enhance the system's adaptability and generalization capabilities.
[0032] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the embodiments disclosed herein. The present application is intended to cover any variations, uses or adaptations of the present application, which follow the general principles of the present application and include common knowledge or customary technical means in the art that are not disclosed in the present application.
[0033] It should be understood that the present application is not limited to the precise structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.
Claims
1. A distribution network section positioning method based on edge collaboration and artificial intelligence, Its characteristics are: The following steps are involved: Step 1: partition the distribution network and set edge nodes; Step 2: When a fault occurs in the distribution network, each node in the partition collects the fault current and uploads it to the edge node of the corresponding partition; Step 3: The edge nodes of each partition aggregate the collected fault current information and share it with the edge nodes of other partitions; Step 4: After receiving the shared data, each edge node constructs a global recognition graph and inputs it into its own fault location model; Step 5: Each edge node independently determines the fault section through its own fault location model and uploads the location result to the voting system; Step 6: The voting system receives the summarized positioning results and performs majority voting to obtain the final fault section; Step 7: The voting system uploads the final fault section location result to the central control system, and the central control system handles the fault.
2. According to claim 1, a distribution network section positioning method based on edge collaboration and artificial intelligence, Its characteristics are: The step 1 is specifically as follows: The distribution network is partitioned by using a distribution network partitioning algorithm based on current transformer data. First, the current data of each node in the distribution network is collected and preprocessed. Then, the correlation matrix between each node is calculated, and the correlation degree of each node in the distribution network is calculated. Construct the Laplace matrix L, perform eigenvalue decomposition on the Laplace matrix L, obtain the eigenvectors corresponding to the first k smallest eigenvalues, form the eigenvector matrix U, use the K-means clustering algorithm to cluster the matrix U, divide the nodes into several sub-segments, and take the cluster center or the point closest to the cluster center as the edge node; According to the clustering results, the segment to which each node belongs is determined, and the distribution network partition is completed.
3. According to claim 2, a distribution network section positioning method based on edge collaboration and artificial intelligence, Its characteristics are: The calculation of the association matrix between the nodes and the degree of association of the nodes in the distribution network are specifically as follows: The current correlation coefficient is used to represent the node correlation. Let the currents of node n and node m be I n (t) and I m (t), correlation coefficient R nm for: In the above formula, and are the average current values of node n and node m in time T respectively; According to the above correlation coefficient R nm , construct the correlation matrix R, the matrix R is an N×N matrix, where the element R nm Represents the association between node n and node m, where N is the number of nodes; The association graph G(V, E) can be obtained from the association matrix, where V is the node in the distribution network and E is the association edge between nodes. nm Exceeding a certain threshold R th When , there is an edge between node n and node m.
4. According to claim 2, a distribution network section positioning method based on edge collaboration and artificial intelligence, Its characteristics are: The expression of the Laplace matrix L is: L=DA Where D is the degree matrix, that is, the diagonal matrix of the number of edges connecting the nodes; A is the adjacency matrix, that is, the matrix composed of the edges between nodes.
5. According to claim 1, a distribution network section positioning method based on edge collaboration and artificial intelligence, Its characteristics are: In the step 2, the collected fault current is also decomposed and denoised, specifically: ICEEMDAN is used to decompose the current signal. First, the original current signal x[n] is taken and Gaussian white noise is added to construct a total of I groups of noisy signals. Let w (i) is Gauss white noise with mean 0 and unit variance; x (i) =x+β 0 E 1 (w (i) ) In the formula, w (i) represents the added i-th group of white noise, where i = 1, 2, ..., I; operator E 1 (·) represents the first IMF component of the EMD decomposition of a signal, β 0 Indicates the weight coefficient when adding noise component, which represents the ratio of the signal-to-noise ratio of the added noise to the standard deviation of the noise component; The local operator M(·) is introduced to represent the local mean of the signal; the local mean is the part obtained by subtracting this IMF from the original signal; Calculate each x (i) The corresponding first-order IMF component Summing the above components, we can get the first-order IMF component D of the zero-sequence current signal x: 1 , The first-order residual r can be obtained 1 , r 1 =x-D 1 The second-order residual r 2 Estimated to be r 1 +β 1 E 2 (w (i) ), The second-order IMF component D can be obtained 2 for Repeat the above steps to get the k-th order residual r k (k=3,4,…,K, K is the total decomposition order) The Kth order IMF component D k for D k =r k-1 -r k Repeat the above steps to obtain the IMF components decomposed by ICEEMDAN.
6. A distribution network section positioning method based on edge collaboration and artificial intelligence according to claim 1, Its characteristics are: The method further includes step 8, feeding back the final fault section location result to each edge node, and each edge node uses the final fault section location result as real data to update the fault location model.
