Cable partial discharge mode identification method based on depth sample enhancement

Abstract

The invention relates to a cable partial discharge mode recognition method based on depth sample enhancement. The method comprises the following steps: converting a collected partial discharge time domain signal of a cable insulation defect into a two-dimensional image, taking the two-dimensional image as an initial sample set, inputting the two-dimensional image into a generative adversarial network, and carrying out the depth sample enhancement, adding the enhanced sample into the initial sample set to form a new sample set, constructing a convolutional neural network model, and performing supervised training on the convolutional neural network model by using the new sample set to obtain a trained convolutional neural network model; converting the time domain signal of the insulation discharge type to be predicted into a two-dimensional image, inputting the two-dimensional image into the trained convolutional neural network model, and outputting the predicted insulation defect discharge type. According to the method, errors caused by feature selection depending on expert experience can be avoided, the insulation defect type of the cable can still be accurately recognized under the condition of small samples, and the recognition rate and accuracy are improved.

Description

technical field [0001] The invention belongs to the technical field of cable partial discharge identification methods, and in particular relates to a cable partial discharge pattern identification method based on depth sample enhancement. Background technique [0002] High-voltage cables are one of the main forms of power transmission, and are widely used in wind farms, solar energy and other distributed energy transmission and grid connection, as well as offshore oil and gas platforms, offshore island power transmission, etc. , Cable body or accessory manufacturing quality, laying installation quality and other reasons cause cable operation failures, which will cause economic losses and social impacts to varying degrees. [0003] The traditional cable partial discharge pattern recognition method relies heavily on the selection of manual features, and the design of manual features is very dependent on expert experience, which will inevitably lead to errors. As an emerging fi...

AI Technical Summary

Problems solved by technology

[0002] High-voltage cables are one of the main forms of power transmission, and are widely used in wind farms, solar energy and other distributed energy transmission and grid connection, as well as offshore oil and gas platforms, offshore island power transmission, etc. , cable body or accessories manufacturing quality, laying installation quality and other reasons caused by cable operation failure will cause different degrees of economic loss and social impact

[0003] The traditional cable partial discharge pattern recognition method relies heavily on the selection of manual features, and the design of manual features is very dependent on expert experience, which will inevitably lead to errors. As an emerging field, deep learning overcomes the above shortcomings and can automatically learn the original The layered representation of data features, but requires a large amount of training data. Therefore, the present invention proposes a method that can avoid errors caused by selecting features based on expert experience, and can still accurately detect cable insulation defects in the case of small samples. Cable Partial Discharge Pattern Recognition Method Based on Depth Sample Enhancement for Type Recognition

Images