Data enhancement method and device for partial discharge spectrum

Abstract

The invention provides a data enhancement method and device for a partial discharge spectrum, and the method comprises the steps: carrying out random disturbance of RGB pixels through employing a Gaussian blur method on the basis of the adding of onsite common noises, such as cellphone interference, radar interference and microwave sulfur lamp interference, so as to train a network with higher generalization capability. A partial discharge spectrum is preprocessed by using background noise coupling, Gaussian blur and other methods, so that one spectrum can generate a plurality of spectrums, alarge number of labeled samples are generated with a small amount of calculation, and the problems of high acquisition cost and insufficient training data of the labeled samples are solved; on the basis of an original data sample, on-site common interference is considered, real on-site data is simulated, and the anti-interference capability of a training model is improved; and the diversity of thetraining data is expanded, the model is trained by the expanded training data, and over-fitting of the model is avoided.

Description

technical field [0001] The present invention relates to the field of partial discharge detection, and more specifically, the present invention relates to a data enhancement method for partial discharge maps. Background technique [0002] With the development of smart grid, the demand for data diagnosis is increasing. Machine learning diagnosis is a solution to replace human diagnosis. When building a machine learning model to train and verify data, how to quickly, efficiently and conveniently obtain a large number of labeled training samples is an urgent problem to be solved. [0003] Partial discharge pattern recognition requires a large number of labeled training samples. The collection process of labeled samples is expensive, and it is often difficult to quickly accumulate enough samples to support training for partial discharge phenomena with low frequency. In traditional partial discharge data enhancement methods, simple processing methods such as translation, rotati...

AI Technical Summary

Problems solved by technology

The graph data processed in these ways: 1) did not take into account the possible interference of data in real business situations; 2) the data enhancement method based on shape processing did not expand the data pixels, and the trained model was only applicable to specified colors and specifications images, poor generalization to data from different sources

Images