State Evaluation System of Switchgear Based on Data Time-domain Processing and Feature Fusion

Abstract

As an important equipment in the power system, it is necessary to detect the running state of high voltage switchgear to ensure its normal operation. But most of current state detection methods for switchgear take a single feature as the basis of fault diagnosis. Failure to make full use of defect information will easily lead to misjudgment and missed detection. In this paper, infrared (IR) and ultraviolet (UV) sensors are used to detect the temperature and partial discharge (PD) information of the insulated equipment in switchgear, which are collected by data acquisition card and transmitted to PC, then, wavelet denoising and sliding window median filtering are carried out on the temperature data to eliminate clutter and spikes on waveform. In order to avoid distortion of pulse width, amplitude and position caused by data processing, a peak-valley extraction and location algorithm is proposed to realize the time domain analysis and feature extraction of PD. And adaptive neuro fuzzy inference strategy (ANFIS) based on Takagi-Sugeno model is used to realize data fusion and state evaluation. Through testing, the accuracy of data processing and feature fusion method in fault warning and diagnosis of test data reaches more than 94%, and the judgment of the running state of electrical equipment is more accurate.