Power distribution network fault diagnosis and positioning method

Abstract

The invention provides a power distribution network fault diagnosis and positioning method. Comprising a power distribution network equal difference image monitoring module, a power distribution network detection data acquisition module, a data fusion processing module and a fault diagnosis and positioning analysis module. The power distribution network equal-difference image monitoring module detects regional power distribution network image information in a graded manner; the power distribution network detection data acquisition module acquires parameter information such as network current and voltage of key nodes; the data fusion processing module processes the regional power distribution network image information and parameter information and then transmits the information to the fault diagnosis and positioning analysis module for storage, the module further comprises a trained fault diagnosis and positioning analysis model, and the model is used for carrying out stepped multi-modal data processing and power distribution network fault diagnosis and positioning. And giving a corresponding early warning level. Real-time fault diagnosis and positioning and early warning reporting can be carried out on the power distribution network, and hidden dangers of the power distribution network are dynamically distinguished, so that a corresponding rapid and accurate fault analysis and maintenance method is given, and the risk and cost of manual inspection are reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of fault diagnosis and optimization, and in particular relates to a method for diagnosing and locating faults in a distribution network. Background Art

[0002] The significance of power supply is very important, as it is related to the national and regional economy, society, energy security and people's lives. Through effective power supply guarantee measures, the normal operation of the power system can be guaranteed, and the reliability and stability of power supply can be ensured, thereby promoting economic development, maintaining social stability and ensuring people's living needs.

[0003] Power lines are the lifeblood of power grid operation. While transmitting electric energy, they are also quite fragile. Any damage from external forces and staff's erroneous operation may cause large-scale power outages, bringing great harm to the social economy and people's lives. The distribution network is directly facing a large number of end users, so its grid structure is complex and the probability of failure is high. A large amount of statistical data shows that 90% of power outages and fault disturbances occur in the distribution network. When a fault occurs, it is necessary to quickly identify and locate the fault point so that timely measures can be taken to deal with it. If the fault cannot be accurately located in the first time, it will lead to the expansion of the power outage scope and the instability of the power system, affecting the quality and safety of users' electricity use, and having an adverse impact on the power supply.

[0004] The rapid diagnosis and location of distribution network line faults is an important part of distribution network operation and maintenance, and plays an important role in rapid power restoration and power supply. At present, the fault diagnosis and location of distribution network lines are still mainly based on manual inspections, supplemented by fault indicator analysis, and generally rely on the operation experience of operation and maintenance personnel, with relatively large uncertainty. In recent years, artificial intelligence has developed rapidly, providing new ideas and new tools for fault diagnosis of distribution network lines.

[0005] The Chinese patent publication number is CN104931857A, and the publication date is September 23, 2015. The name of the invention is a distribution network fault location method based on D□S evidence theory. The scheme discloses a distribution network fault location method based on DS evidence theory. Its shortcomings are that the distribution network has many branches and the grounding points are hidden, resulting in a long time for locating the grounding fault, and the accuracy of locating the grounding line fault point is not high.

[0006] It can be seen that the existing technology has the following problems: the existing model processing distribution network fault location method does not process the distribution network data by level and region, and the algorithm model used is not well applicable to the distribution network with multiple transformers and substations. In addition, the image data of the node parameter points of the regional distribution network is not well utilized, and the algorithm model used cannot change dynamically to obtain accurate fault identification and regional location.

[0007] Therefore, there is an urgent need for a method for distribution network fault diagnosis and location to obtain and process image data of specific nodes for modular fault area identification, and to combine the distribution network node networking method to diagnose and locate existing distribution network faults. Summary of the invention

[0008] In order to solve the above technical problems, the present invention proposes a method for diagnosing and locating distribution network faults.

[0009] In a first aspect of the present invention, a method for diagnosing and locating a distribution network fault is provided, the method comprising:

[0010] A. Obtain the first-level regional distribution network image, record the first key node parameter information at this time, and obtain the first historical fault cause and area from the distribution network historical data;

[0011] B. Processing the first-level regional distribution network image and the first key node parameter information to generate first fault diagnosis and positioning parameters;

[0012] C. Generate a fault diagnosis and location analysis model based on the first fault diagnosis and location parameter and the first historical fault cause and region, the fault diagnosis and location analysis model comprehensively analyzes the key node parameters of the distribution network and the hierarchical regional network image, and performs fault diagnosis, location and early warning reporting on the distribution network;

[0013] D. Obtaining a second-level regional distribution network image and recording the second key node parameter information at this time, wherein the second-level regional distribution network image and the second key node parameter information constitute a second fault diagnosis and positioning parameter;

[0014] E. Use the fault diagnosis and location analysis model in step C to process the second fault diagnosis and location parameters to generate a second real-time fault cause and area, and perform fault diagnosis, location and early warning reporting on the current distribution network based on the second real-time fault cause and area.

[0015] Furthermore, the first key node parameter information and the second key node parameter information include the current change rate at the distribution network transformer, the voltage average value at the nodes of the entire distribution network, and the average value of the node degree after the entire distribution network is networked; the first-level regional distribution network image and the second-level regional distribution network image are the pixel average values ​​of the processed entire distribution network image.

[0016] Furthermore, the fault diagnosis and location analysis model is constructed using a neural network or support vector machine method; the first-level regional distribution network image and the second-level regional distribution network image are obtained by shooting in graded time periods using cameras fixed to transformers or substations.

[0017] Furthermore, the first-level regional distribution network image and the second-level regional distribution network image are for grading the distribution network area levels of the distribution network, which are divided into a power generation end distribution network area image, a transmission transformer end distribution network area image, a distribution substation end distribution network area image, and a transformer end distribution network area image. The power generation end distribution network area image is the distribution network image between the transmission substation and the transmission plant, which is a first-level distribution network image; the transmission transformer end distribution network area image is the distribution network image between the transmission substation and the distribution substation, which is a second-level distribution network image; the distribution substation end distribution network area image is the distribution network image between the distribution substation and the transformer, which is a third-level distribution network image; the transformer end distribution network area image is the distribution network image from the transformer to the residential user end, which is a fourth-level distribution network image.

[0018] Furthermore, before using the neural network or support vector machine method, the first-level regional distribution network image and the second-level regional distribution network image are processed by using improved multilateral noise removal processing.

[0019] Furthermore, before using the deep learning or machine learning method, the authenticity of the first-level regional distribution network image and the second-level regional distribution network image is confirmed.

[0020] Furthermore, the Rt parameter of the image set to be fused and the corresponding first key node parameter information or the second key node parameter information are input into the support vector machine model in a combined form, and the cause and area of ​​the distribution network fault are output. The combined form is specifically the grayscale value of the image to be fused and the current change rate at the distribution network transformer, the voltage average value at the node of the entire distribution network, and the average value of the node degree after the entire distribution network is networked.

[0021] Furthermore, the support vector machine adopts a support vector machine model improved according to the hierarchical regional distribution network image.

[0022] A control device for distribution network fault diagnosis and location is also provided, comprising a memory and a processor, wherein a control program for distribution network fault diagnosis and location is stored in the memory, and the processor is configured to run the control program for distribution network fault diagnosis and location to run a method for distribution network fault diagnosis and location.

[0023] A computer-readable storage medium is also provided, characterized in that it includes: computer-executable instructions, which, when executed, execute a method for diagnosing and locating distribution network faults.

[0024] The present invention uses a trained historical model to obtain real-time fault area information in a timely manner. Before the model is trained to process the corresponding data, it also processes the distribution network data by level and region. The algorithm model used is well applicable to the current complex multi-transformer and substation distribution network. It also makes good use of the image data of the node parameter points of the regional distribution network, and the algorithm model used can be dynamically changed in real time according to the number of transformers in different distribution networks to obtain accurate fault identification and regional positioning.

[0025] More embodiments and improved effects of the present invention will be further introduced in conjunction with the drawings and specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is a flow chart of a method for diagnosing and locating faults in a distribution network of the present invention;

[0027] Figure 2 is a partial schematic diagram of the power distribution network of the present invention;

[0028] Figure 3 is a schematic diagram of the comparison image before and after noise removal of the present invention;

[0029] Figure 4 is a schematic diagram of fault area diagnosis of a distribution network of the present invention;

[0030] Figure 5 is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention; DETAILED DESCRIPTION

[0031] The invention is further described below in conjunction with the accompanying drawings and specific implementation methods.

[0032] In order to solve the above technical problems, the present invention proposes a method for diagnosing and locating distribution network faults.

[0033] In a first aspect of the present invention, a method for diagnosing and locating a distribution network fault is provided, the method comprising:

[0034] A. Obtain the first-level regional distribution network image, record the first key node parameter information at this time, and obtain the first historical fault cause and area from the distribution network historical data;

[0035] B. Processing the first-level regional distribution network image and the first key node parameter information to generate first fault diagnosis and positioning parameters;

[0036] C. Generate a fault diagnosis and location analysis model based on the first fault diagnosis and location parameter and the first historical fault cause and region, the fault diagnosis and location analysis model comprehensively analyzes the key node parameters of the distribution network and the hierarchical regional network image, and performs fault diagnosis, location and early warning reporting on the distribution network;

[0037] D. Obtaining a second-level regional distribution network image and recording the second key node parameter information at this time, wherein the second-level regional distribution network image and the second key node parameter information constitute a second fault diagnosis and positioning parameter;

[0038] E. Use the fault diagnosis and location analysis model in step C to process the second fault diagnosis and location parameters to generate a second real-time fault cause and area, and perform fault diagnosis, location and early warning reporting on the current distribution network based on the second real-time fault cause and area.

[0039] Furthermore, the first key node parameter information and the second key node parameter information include the current change rate I at the distribution network transformer, the voltage average value U at the nodes of the entire distribution network, and the average value Y of the node degree after the entire distribution network is networked; the first-level regional distribution network image and the second-level regional distribution network image are the pixel average values ​​of the processed entire distribution network image.

[0040] Furthermore, the fault diagnosis and location analysis model is constructed using a neural network or support vector machine method; the first-level regional distribution network image and the second-level regional distribution network image are obtained by shooting in graded time periods using cameras fixed around the transformer or substation.

[0041] Furthermore, the first-level regional distribution network image and the second-level regional distribution network image are for grading the distribution network area levels of the distribution network, which are divided into a power generation end distribution network area image, a transmission transformer end distribution network area image, a distribution substation end distribution network area image, and a transformer end distribution network area image. The power generation end distribution network area image is the distribution network image between the transmission substation and the transmission plant, which is a first-level distribution network image; the transmission transformer end distribution network area image is the distribution network image between the transmission substation and the distribution substation, which is a second-level distribution network image; the distribution substation end distribution network area image is the distribution network image between the distribution substation and the transformer, which is a third-level distribution network image; the transformer end distribution network area image is the distribution network image from the transformer to the residential user end, which is a fourth-level distribution network image.

[0042] In this embodiment, the first-level regional power distribution network image and the second-level regional power distribution network image are obtained by averaging four-level power distribution network images.

[0043] Furthermore, before using the neural network or support vector machine method, the first-level regional distribution network image and the second-level regional distribution network image are processed, and the improved multilateral denoising process is adopted, and the multilateral denoising image I(G) is specifically obtained as follows:

[0044]

[0045] Where I(G) represents the result after multilateral denoising, L represents the number of transmission substations in the distribution network, K represents the number of distribution substations in the distribution network, B represents the number of transformers in the distribution network, S represents the location scale parameter, P represents the grayscale scale parameter, and x i and x j Represents different positions of a two-dimensional pixel image, I i and I j Represents the grayscale values ​​of different pixels in a two-dimensional pixel image, K x Represents a normalizing constant.

[0046] In this embodiment, the multilateral denoising image is obtained by taking the average grayscale value of the four-level distribution network images after performing multilateral denoising on the four-level distribution network images.

[0047] Furthermore, before using the deep learning or machine learning method, the authenticity of the first-level regional distribution network image and the second-level regional distribution network image is confirmed, and the image true value R (d, f) is:

[0048]

[0049] In the formula, E d and E f They represent the first-level regional distribution network image and the second-level regional distribution network image and the mean of their filtered images, respectively, and f d and f f They represent the variances of the first-level regional distribution network image and the second-level regional distribution network image and their filtered images, respectively, and f df It represents the covariance of the image, which reflects the gray value information, contrast information and structural similarity information of the image. g1 and g2 are two constants close to zero. After confirming the true value of the image, the images with true values ​​less than the threshold are removed to obtain the image set Rt to be fused.

[0050] In this embodiment, the true value R is used to remove the severely distorted image after noise removal.

[0051] Furthermore, the Rt parameter of the image set to be fused and the corresponding first key node parameter information or the second key node parameter information are combined and input into the support vector machine model, and the cause and area of ​​the distribution network fault are output. The combination form is specifically the gray value of the image to be fused and the current change rate at the distribution network transformer, the voltage average value at the node of the entire distribution network, and the average value of the node degree after the entire distribution network is networked. In this embodiment, the combined fault diagnosis and positioning parameter characteristic value is (G, I, U, Y), and the parameter is input into the support vector machine model to obtain the corresponding fault area. The fault area is shown in Figure 2 or Figure 4 As shown, it is divided into four areas, namely, the distribution network area image area at the power generation end, the distribution network area image area at the transmission transformer end, the distribution network area image area at the distribution substation end, and the distribution network area image area at the transformer end. The distribution network area image at the power generation end is the distribution network image area between the transmission substation and the transmission plant, which is the first-level distribution network image area. The distribution network area image at the transmission transformer end is the distribution network image area between the transmission substation and the distribution substation, which is the second-level distribution network image area. The distribution network area image at the distribution substation end is the distribution network image area between the distribution substation and the transformer, which is the third-level distribution network image area. The distribution network area image at the transformer end is the distribution network image area between the transformer and the residential user end, which is the fourth-level distribution network image area.

[0052] Furthermore, the support vector machine adopts a support vector machine model improved according to the hierarchical regional distribution network image, and its specific function expression is:

[0053]

[0054] Where ω, b are the normal vector and intercept of the hyperplane, respectively, X i is an input variable, which is obtained by combining the features of the current change rate at the distribution network transformer in the training data image set Rt corresponding to the image acquisition time period, the average voltage at the node of the entire distribution network, and the average node degree of the entire distribution network after networking. i is the number of samples, D d and D f They represent the grayscale mean of the hierarchical regional distribution network image and the denoised image respectively, N represents the number of transformer nodes after networking, and M represents the number of nodes of the transmission substation, distribution substation, and transformer after networking.

[0055] In this embodiment, see Figure 2 As shown, N is 2 and M is 5. If the value of the classifier is greater than If the value of the function expression is greater than 1000, it means that there is a bleeding fault in the image area of ​​the fourth-level distribution network.

[0056] A control device for distribution network fault diagnosis and location is also provided, comprising a memory and a processor, wherein a control program for distribution network fault diagnosis and location is stored in the memory, and the processor is configured to run the control program for distribution network fault diagnosis and location to run a method for distribution network fault diagnosis and location.

[0057] A computer-readable storage medium is also provided, characterized in that it includes: computer-executable instructions, which, when executed, execute a method for diagnosing and locating distribution network faults.

[0058] A system for fault diagnosis and location of distribution network is also provided. It includes a distribution network equidistant image monitoring module, a distribution network detection data acquisition module, a data fusion processing module and a fault diagnosis and location analysis module. The distribution network equidistant image monitoring module detects regional distribution network image information in different levels, and the distribution network detection data acquisition module collects parameter information such as network current and voltage of key nodes; the data fusion processing module processes the regional distribution network image information and parameter information and transmits them to the fault diagnosis and location analysis module for storage. The module also includes a trained fault diagnosis and location analysis model, which uses the model to perform step-by-step multimodal data processing and distribution network fault diagnosis and location, and gives corresponding warning levels. It is capable of performing real-time fault diagnosis, location and warning reports on the distribution network, dynamically identifying hidden dangers in the power grid, and thus giving corresponding fast and accurate fault analysis and maintenance methods, reducing the risk and cost of manual inspections.

[0059] The present invention uses a trained historical model to obtain real-time fault area information in a timely manner. Before training and processing the corresponding data, the model also processes the distribution network data by level and region. The algorithm model used is well applicable to the current complex multi-transformer and substation distribution network. It also makes good use of the image data of the node parameter points of the regional distribution network, and the algorithm model used can be dynamically changed in real time according to the number of transformers in different distribution networks to obtain accurate fault identification and regional positioning.

[0060] Of course, it can be understood that each embodiment of the present invention can achieve one of the effects alone, and a combination of multiple embodiments of the present invention can achieve all of the above effects, but it is not required that every embodiment of the present invention achieve all of the above advantages and effects, because each embodiment of the present invention can constitute a separate technical solution and make one or more contributions to the prior art.

[0061] The present invention does not particularly specify the structure of some modules, which shall be subject to the contents recorded in the prior art. The prior art mentioned in the above background technology section and the specific embodiment section of the present invention can be used as part of the present invention to understand the meaning of some technical features or parameters. The scope of protection of the present invention shall be subject to the contents actually recorded in the claims.

Claims

1. A method for diagnosing and locating distribution network faults, It is characterized in that The method comprises: A. Obtain the first-level regional distribution network image, record the first key node parameter information at this time, and obtain the first historical fault cause and area from the distribution network historical data; B. Processing the first-level regional distribution network image and the first key node parameter information to generate first fault diagnosis and positioning parameters; C. Generate a fault diagnosis and location analysis model based on the first fault diagnosis and location parameter and the first historical fault cause and region, the fault diagnosis and location analysis model comprehensively analyzes the key node parameters of the distribution network and the hierarchical regional network image, and performs fault diagnosis, location and early warning reporting on the distribution network; D. Obtaining a second-level regional distribution network image and recording the second key node parameter information at this time, wherein the second-level regional distribution network image and the second key node parameter information constitute a second fault diagnosis and positioning parameter; E. Use the fault diagnosis and location analysis model in step C to process the second fault diagnosis and location parameters to generate a second real-time fault cause and area, and perform fault diagnosis, location and early warning reporting on the current distribution network based on the second real-time fault cause and area.

2. A method for diagnosing and locating distribution network faults as claimed in claim 1, Features: The first key node parameter information and the second key node parameter information include the current change rate at the distribution network transformer, the voltage average value at the nodes of the entire distribution network, and the average value of the node degree after the entire distribution network is networked; the first-level regional distribution network image and the second-level regional distribution network image are the pixel average values ​​of the processed entire distribution network image.

3. A method for diagnosing and locating distribution network faults as claimed in claim 2, Features: The fault diagnosis and location analysis model is constructed using a neural network or support vector machine method; the first-level regional distribution network image and the second-level regional distribution network image are obtained by shooting in graded time periods using cameras fixed around transformers or substations.

4. A method for diagnosing and locating distribution network faults as claimed in claim 1, Features: The first-level regional distribution network image and the second-level regional distribution network image are for grading the distribution network area levels of the distribution network, which are divided into a power generation end distribution network area image, a transmission transformer end distribution network area image, a distribution substation end distribution network area image, and a transformer end distribution network area image. The power generation end distribution network area image is the distribution network image between the transmission substation and the transmission plant, which is a first-level distribution network image; the transmission transformer end distribution network area image is the distribution network image between the transmission substation and the distribution substation, which is a second-level distribution network image; the distribution substation end distribution network area image is the distribution network image between the distribution substation and the transformer, which is a third-level distribution network image; the transformer end distribution network area image is the distribution network image from the transformer to the residential user end, which is a fourth-level distribution network image.

5. A method for diagnosing and locating distribution network faults as claimed in claim 4, Features: Before using the neural network or support vector machine method, the first-level regional distribution network image and the second-level regional distribution network image are processed by using improved multilateral noise removal processing.

6. A method for diagnosing and locating distribution network faults as claimed in claim 5, Features: Before using deep learning or machine learning methods, the authenticity of the first-level regional distribution network image and the second-level regional distribution network image is confirmed.

7. A method for diagnosing and locating distribution network faults as claimed in claim 6, Features: The Rt parameter of the image set to be fused and the corresponding first key node parameter information or the second key node parameter information are combined and input into the support vector machine model, and the cause and area of ​​the distribution network fault are output. The combination form is specifically the grayscale value of the image to be fused and the current change rate at the distribution network transformer, the voltage average value at the node of the entire distribution network, and the average value of the node degree after the entire distribution network is networked.

8. A method for diagnosing and locating distribution network faults as claimed in claim 3, Features: The support vector machine adopts a support vector machine model improved according to the hierarchical regional distribution network image.

9. A control device for diagnosing and locating faults in a distribution network, comprising a memory and a processor, It is characterized in that The memory stores a control program for distribution network fault diagnosis and location, and the processor is configured to run the control program for distribution network fault diagnosis and location to run the method described in any one of claims 1 to 8.

10. A computer-readable storage medium, It is characterized in that The method comprises: computer executable instructions, which, when executed, execute the method according to any one of claims 1 to 8.

Images