A power distribution network cable joint defect detection method and system and a storage medium

Abstract

The application discloses a kind of power distribution network cable joint defect detection method, system and storage medium, it is related to power distribution network safety field.The present application includes the following steps: extracting the characteristic data of power distribution network cable joint, wherein the characteristic data includes temperature characteristic parameter, partial discharge characteristic parameter and sound characteristic parameter;Characteristic data is normalized and input into power distribution network cable joint defect detection model;Identify output results, determine defect type.The present application uses the characteristic data parameter of power distribution network cable joint to establish model, and uses model to identify defect, again in combination with the mapping relationship between partial discharge characteristic parameter and defect, determine defect type, improve the identification ability of cable joint defect.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of power distribution network safety, and more particularly to a power distribution network cable joint defect detection method and system and a storage medium. BACKGROUND

[0002] As the main equipment of the urban power transmission and distribution network, the safety and reliability of the power cable are of great significance to the normal operation of the power system. With the long-term operation of the cable, the insulation aging or the increase of the contact resistance caused by the cable joint defect will cause the overall temperature of the joint to rise, and the rise in temperature will positively promote the aging of the insulation material. If the change of the joint temperature cannot be detected in time, it will endanger the safe operation of the power transmission and distribution network.

[0003] Compared with the cable body, the cable joint is prone to cause main insulation defects in the manufacturing and installation process due to its special and complex structure. The defects caused in the manufacturing and installation process of the cable joint mainly include air gap defects, moisture defects, sharp end defects, impurity defects, etc. With the long-term operation of the cable, these defects with different material properties will cause the accumulation of space charge in the joint and cause the distortion of the electric field of the main insulation part, thereby causing local discharge, temperature change and other phenomena. Further, with the long-term operation of the cable, the existence of the defects will gradually cause the deterioration and failure of the insulation performance, and further cause the occurrence of power accidents. Therefore, accurately and timely detecting and diagnosing the defects of the cable joint is of great significance to the later maintenance of the equipment and the stable operation of the power distribution network. At present, the diagnosis method of the cable joint defect is usually to extract features and recognize patterns of a single discharge signal. Influenced by the detection environment, human operation and other factors, the recognition rate of the diagnosis result is not high enough, and the rich insulation state information cannot be fully excavated. SUMMARY

[0004] Therefore, the present application provides a power distribution network cable joint defect detection method, system and storage medium to solve the problems in the background art.

[0005] In order to achieve the above purpose, the present application adopts the following technical scheme:

[0006] A power distribution network cable joint defect detection method, comprising the following steps:

[0007] extracting feature data of the power distribution network cable joint, wherein the feature data includes temperature characteristic parameters, local discharge characteristic parameters and sound characteristic parameters;

[0008] normalizing the feature data and inputting it into a power distribution network cable joint defect detection model;

[0009] recognizing the output result to determine the defect type.

[0010] Optionally, the power distribution network cable joint defect detection model comprises a temperature feature parameter detection part, a discharge feature parameter detection part, a sound feature parameter detection part, and a fusion part; the temperature feature parameter detection part, the discharge feature parameter detection part, and the sound feature parameter detection part are all input to the fusion part for fusion processing and output of a detection result.

[0011] Optionally, when acquiring the temperature feature parameter, the thermal network method is used to subdivide the temperature object to be measured of the power distribution network cable joint into a plurality of unit nodes, and the internal temperature of each node is preset to be uniform, the heat transfer between the nodes is preset to be thermal resistance conduction, and the temperature feature parameter is determined by using the law of conservation of heat.

[0012] Optionally, the temperature feature parameter detection part, the discharge feature parameter detection part, and the sound feature parameter detection part perform normalization processing on the temperature feature parameter, the partial discharge feature parameter, and the sound feature parameter respectively, and input the normalization processing result to the corresponding neural network part; the neural network part extracts multi-scale space-time features of temperature, discharge, and sound through a mixed multi-scale feature extraction network; the fusion part extracts complementary information in the space-time features based on a feature interaction module of Non-local operation to realize the interaction of the three parameter features; and then the feature sharing module and a sample selection strategy based on a three-tuple loss function are used to realize the clustering and fusion of the space-time features in a high-dimensional embedding space, the fusion result is judged based on a vector machine, and a detection result 0 or 1 is output, where 0 represents no defect and 1 represents a defect.

[0013] Optionally, the output result is identified to determine the defect type, specifically: deep learning is performed on the feature data with the output result of 1, and a defect type is output according to the deep learning result.

[0014] Optionally, it further comprises a wavelet threshold denoising method for denoising the feature data.

[0015] A power distribution network cable joint defect detection system comprises the following steps:

[0016] A feature data extraction module is used to extract feature data of the power distribution network cable joint, wherein the feature data comprises a temperature feature parameter, a partial discharge feature parameter, and a sound feature parameter.

[0017] A defect detection module is used to perform normalization processing on the feature data and input the feature data into a power distribution network cable joint defect detection model.

[0018] A defect identification module is used to identify the output result and determine the defect type.

[0019] A computer storage medium, the computer storage medium has a computer program stored thereon, the computer program is executed by a processor to realize the steps of any one of the power distribution network cable joint defect detection method.

[0020] Through the above technical solution, compared with the prior art, the power distribution network cable joint defect detection method, system and storage medium provided by the application have the following beneficial effects:

[0021] 1. A model is established using characteristic data parameters of the power distribution network cable joint, and the model is used to identify defects, and the mapping relationship between the local discharge characteristic parameters and the defects is combined again to determine the defect type, thereby improving the recognition ability of the cable joint defects.

[0022] 2. The defect detection model is used to identify typical defects of the cable joint based on three characteristic data (temperature characteristic parameters, local discharge characteristic parameters and sound characteristic parameters), and the recognition rate is high. Tests show that the method has high reliability. BRIEF DESCRIPTION OF DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are only a part of the embodiments of the application, and for those skilled in the art, other drawings can be obtained without creative labor on the basis of the provided drawings.

[0024] Figure 1 The flowchart of the application is shown in the figure.

[0025] Figure 2 The structural diagram of the application is shown in the figure. DETAILED DESCRIPTION

[0026] The technical solutions in the embodiments of the application will be described clearly and completely in the following with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only a part of the embodiments of the application, not all. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the application.

[0027] The embodiment of the application discloses a power distribution network cable joint defect detection method, as shown in the figure, comprising the following steps: Figure 1

[0028] S1: Extracting characteristic data of the power distribution network cable joint, wherein the characteristic data includes temperature characteristic parameters, local discharge characteristic parameters and sound characteristic parameters;

[0029] ​S2: Normalize the feature data and input it into the power distribution network cable joint defect detection model;

[0030] S3: Recognize the output result and determine the defect type.

[0031] Further, in S1, the temperature feature parameter is obtained by using the heat network method to subdivide the temperature object to be measured of the power distribution network cable joint into a plurality of unit nodes, and the internal temperature of the node is assumed to be uniform, the heat transfer between the nodes is assumed to be thermal resistance conduction, and the temperature feature parameter is determined by using the law of conservation of heat. Specifically, the heat network method subdivides the research object into a plurality of unit nodes, and assumes that each node is a unit with lumped parameters, i.e. the internal temperature of the node is assumed to be uniform, and the heat transfer between the nodes, whether it is thermal conduction, thermal convection or thermal radiation, is connected by thermal resistance, and further forms a thermal network. The node heat flow includes the self-generated heat of the node, the external heat load, and the heat transfer between the nodes. The thermal network model introduces the concepts of thermal resistance and thermal capacity, and determines the temperature feature parameter by means of the form of Kirchhoff's law.

[0032] Further, in S1, the partial discharge feature parameter is obtained by PRPD pattern recognition, and the phase distribution of different partial discharge types has certain differences. By statistically analyzing the relationship among the power frequency phase, the number of discharges n and the discharge amount q, a statistical element based on the phase distribution is extracted as the basis for judging the discharge type.

[0033] Further, in S2, the power distribution network cable joint defect detection model includes a temperature feature parameter detection part, a discharge feature parameter detection part, a sound feature parameter detection part and a fusion part; the temperature feature parameter detection part, the discharge feature parameter detection part and the sound feature parameter detection part are all input into the fusion part for fusion processing and output of the detection result.

[0034] Further, the temperature feature parameter detection part, the discharge feature parameter detection part and the sound feature parameter detection part respectively normalize the temperature feature parameter, the partial discharge feature parameter and the sound feature parameter, and input the normalized processing result into the corresponding neural network part; the neural network part extracts the multi-scale space-time features of temperature, discharge and sound through a mixed multi-scale feature extraction network; the fusion part extracts complementary information in the space-time features based on a feature interaction module of Non-local operation, realizes the interaction of the three parameter features, and then realizes the clustering and fusion of the space-time features in the high-dimensional embedding space through a feature sharing module and a sample selection strategy based on a three-tuple loss function. The fusion result is judged based on a vector machine, and the detection result 0 or 1 is output, 0 indicating that there is no defect, and 1 indicating that there is a defect.

[0035] Furthermore, this invention employs depthwise separable convolutions instead of standard convolutions to construct the basic residual units of the network with a residual mechanism, and captures multi-scale information of the input features by stacking multiple residual units. The network's basic module—the multi-scale module—contains four convolutional branches at different scales. The i-th convolutional branch is composed of stacked residual units and is used to capture features at scale 1. To adaptively aggregate the acquired features at different scales, this invention adds a feature aggregation component composed of shared fully connected (FC) units at the end of the multi-scale module. Depending on the features, the feature aggregation component assigns specific weights to fuse features at different scales.

[0036] Furthermore, in S3, identifying the output result and determining the defect type specifically involves: performing deep learning on the feature data with an output result of 1, and outputting the defect type based on the deep learning result.

[0037] The present invention also includes a method for denoising feature data using wavelet thresholding.

[0038] Specifically, the acquired partial discharge signals are often mixed with external interference signals. After denoising, the measured signal is made as close as possible to the true value of the ideal discharge signal. Wavelet threshold denoising is advantageous among many denoising algorithms because it can preserve the edges and discontinuities of the original waveform to the greatest extent. Its basic idea is to set a threshold for the signal to be processed after wavelet transform. If the wavelet coefficients are higher than the threshold, the part is defined as a valid signal that can be retained; if the wavelet coefficients are lower than the threshold, it is defined as noise that needs to be filtered out. The specific steps are as follows:

[0039] Select an appropriate wavelet and decomposition level to complete the wavelet transform of the signal to be processed, and obtain the wavelet decomposition coefficients according to the set threshold function or threshold.

[0040] The wavelet decomposition coefficients are denoised to obtain the processed wavelet coefficients.

[0041] The wavelet coefficients after thresholding are subjected to inverse wavelet transform to obtain the denoised signal.

[0042] and Figure 1 Corresponding to the method shown, the present invention also discloses a distribution network cable joint defect detection system for... Figure 1 The implementation of the method, specifically its structure, is as follows: Figure 2 As shown, it includes the following steps:

[0043] Feature data extraction module: used to extract feature data of cable joints in power distribution networks, including temperature feature parameters, partial discharge feature parameters, and sound feature parameters;

[0044] The defect detection module is configured to normalize the feature data and input the normalized feature data into the power distribution network cable joint defect detection model.

[0045] The defect recognition module is configured to recognize the output result and determine the defect type.

[0046] The embodiment further discloses a computer storage medium, and the computer storage medium stores a computer program. The computer program is executed by a processor to implement the steps of any one of the power distribution network cable joint defect detection methods.

[0047] The embodiments in the specification are described in a progressive manner, and each embodiment focuses on the difference from other embodiments. The same or similar parts of each embodiment can be referred to each other. For the device disclosed by the embodiments, since it corresponds to the method disclosed by the embodiments, the description is relatively simple, and the related parts can be referred to the method part.

[0048] The above description of the disclosed embodiments enables a person skilled in the art to implement or use the present application. Various modifications of the embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for detecting defects in cable joints in power distribution networks, characterized in that, Includes the following steps: Extract the characteristic data of cable joints in the power distribution network, including temperature characteristic parameters, partial discharge characteristic parameters, and sound characteristic parameters. The feature data is normalized and then input into the power distribution network cable joint defect detection model; Identify the output results and determine the defect type; The defect detection model for power distribution network cable joints includes a temperature characteristic parameter detection part, a discharge characteristic parameter detection part, a sound characteristic parameter detection part, and a fusion part; the temperature characteristic parameter detection part, the discharge characteristic parameter detection part, and the sound characteristic parameter detection part are all input to the fusion part for fusion processing and output detection results; The temperature feature parameter detection part, the discharge feature parameter detection part, and the sound feature parameter detection part normalize the temperature feature parameter, the partial discharge feature parameter, and the sound feature parameter, respectively, and input the normalization result into the corresponding neural network part. The neural network part extracts multi-scale spatiotemporal features of temperature, discharge, and sound through a hybrid multi-scale feature extraction network. The fusion part extracts complementary information in the spatiotemporal features based on a non-local operation feature interaction module to realize the interaction of the three parameter features. Then, the spatiotemporal features are clustered and fused in the high-dimensional embedding space through a feature sharing module and a sample selection strategy based on a triplet loss function. The fusion result is judged based on a vector machine, and the detection result is output as 0 or 1, where 0 indicates no defect and 1 indicates defect.

2. The method for detecting defects in cable joints in a power distribution network according to claim 1, characterized in that, When obtaining temperature characteristic parameters, the thermal network method is used to subdivide the temperature target of the distribution network cable joint into several unit nodes, and the internal temperature of the nodes is assumed to be uniform, and the heat transfer between the nodes is assumed to be thermal resistance conduction. The temperature characteristic parameters are determined by using the law of conservation of heat.

3. The method for detecting defects in cable joints in a power distribution network according to claim 1, characterized in that, The specific steps for identifying the output result and determining the defect type are as follows: perform deep learning on the feature data with an output result of 1, and output the defect type based on the deep learning result.

4. The method for detecting defects in cable joints in a power distribution network according to claim 1, characterized in that, It also includes denoising the feature data using wavelet thresholding.

5. A defect detection system for cable joints in power distribution networks, characterized in that, Includes the following steps: Feature data extraction module: used to extract feature data of cable joints in power distribution networks, including temperature feature parameters, partial discharge feature parameters, and sound feature parameters; Defect detection module: used to normalize feature data and input it into the distribution network cable joint defect detection model; the distribution network cable joint defect detection model includes a temperature feature parameter detection part, a discharge feature parameter detection part, a sound feature parameter detection part, and a fusion part; the temperature feature parameter detection part, the discharge feature parameter detection part, and the sound feature parameter detection part are all input into the fusion part for fusion processing and output the detection results; The temperature feature parameter detection section, the discharge feature parameter detection section, and the sound feature parameter detection section normalize the temperature feature parameter, the partial discharge feature parameter, and the sound feature parameter, respectively, and input the normalization result into the corresponding neural network section. The neural network section extracts multi-scale spatiotemporal features of temperature, discharge, and sound through a hybrid multi-scale feature extraction network. The fusion section extracts complementary information in the spatiotemporal features based on a non-local operation feature interaction module to realize the interaction of the three parameter features. Then, the spatiotemporal features are clustered and fused in the high-dimensional embedding space through a feature sharing module and a sample selection strategy based on a triplet loss function. The fusion result is judged based on a vector machine, and the detection result is output as 0 or 1, where 0 indicates no defect and 1 indicates defect. Defect identification module: Used to identify the output results and determine the defect type.

6. A computer storage medium, characterized in that, The computer storage medium stores a computer program, which, when executed by a processor, implements the steps of a method for detecting defects in power distribution network cable joints as described in any one of claims 1-4.

Images