A Separation and Classification Method for Discharge Signals of Power Equipment Based on Kernel Principal Component Analysis

Abstract

The invention discloses a method for separating and classifying discharge signals of electric equipment based on nuclear principal component analysis, which comprises the steps of: (1) acquiring characteristic quantities of electric equipment discharge signals as a sample data set; (2) processing the sample data set with nuclear principal component analysis , the obtained principal component variables are used as new feature quantities; (3) cluster the discharge signals through the K-means clustering method to obtain the type labels of different types of discharge signals; (4) select the Principal component variables, drawing two-dimensional and three-dimensional scatter diagrams; (5) Gates between different types of discharge signals are performed according to the type labels to complete the automatic classification of discharge signals. The method of the present invention adopts a nonlinear method to extract the principal component features, selects a suitable kernel function to extract the principal components of the discharge signal from the original discharge pulse signal, and realizes the automatic separation and classification of multiple discharge source pulses and multiple interference source pulses, thereby realizing the discharge of electric equipment type identification.

Description

technical field [0001] The invention relates to the field of separation and classification of discharge signals of electric power equipment, in particular to a method for separation and classification of discharge signals of electric power equipment based on nuclear principal component analysis. Background technique [0002] Insulation faults of high-voltage power equipment are often caused by penetrating discharges inside or on the surface of insulating materials, and penetrating discharge channels are generally developed from tiny discharge signals. The weak discharge current pulse at the initial stage of discharge is also far lower than the stray pulses in the power system, and is easily overwhelmed by background noise or corona interference signals. [0003] Based on the characteristics of the discharge signal of power equipment, if the discharge signal cannot be separated from various background spurious signals, it is impossible to identify and classify the fault type ...

AI Technical Summary

Problems solved by technology

Conventional discharge signal separation and classification methods are usually based on the pulse phase distribution spectrum (PRPD) and pulse waveform statistical characteristic parameters to identify a single discharge type, but the actual field operation of power equipment is affected by the harsh operating environment on site, and there are multiple discharges at the same time. The possibility of discharge types, the traditional separation and classification methods cannot identify multiple discharge types and multiple interference signal sources

Images