Partial Discharge Pattern Identification Based on Its Non-linear Dynamic Characteristics

Abstract

Partial discharge (PD) on-line monitoring can detect the insulated defect of electric equipment effectively, which will prevent suddenly accident. On the basis of PD mechanism research, the relation between PD detecting signal non-linear dynamics eigenvector and insulation defect pattern is investigated by experiment. Using the characteristic parameters, such as Lyapunov exponent, Kolmogorov, fractal dimension extracted from the PD detecting signal time sequence data to be the eigenvector of PD pattern identification. A series of typical insulation PD experiments is carried out in the lab. Through analyzing the non-linear dynamics character of PD detecting signal time sequence, such as Lyapunov exponent, fractal dimension, Kolmogorov entropy, the PD process is identified be a non-linear chaotic system under some voltage condition. A series of dynamics estate equations have been set up and analyzed, to go along PD pattern identification. The typical insulation defect models used to the PD experiment include needle-broad poles discharge of oil, insulation cardboard interior gas discharge, cardboard surface discharge, suspending discharge in oil and so on. Broadband PD detecting system and digital memorial oscillograph TDS3054B are used to measure PD impulse current signal and ultrasonic signal. The experiment diagram and experiment result are presented in this paper, which shows this method has a favorable PD pattern identification effect. It's significant to the insulation condition-based maintenance of high-voltage power equipment.