High Accuracy Insulation Fault Diagnosis Method of Power Equipment Based on Power Maximum Likelihood Estimation

Abstract

Partial discharge (PD) detection is one of the most effective methods for electrical equipment insulation fault diagnosis. Most existing PD detection methods are based on the processing of PD waveforms. However, the PD waveforms can be severely disturbed by the complex environment noise in substations, which can lead to great drop in detection accuracy. In this paper, we propose a PD detection method based on power maximum likelihood estimation. This method does not rely on the PD waveforms but utilizes a statistical approach of maximum likelihood estimation to analyze the distribution characteristic of PD signals, transferring the traditional method of waveform processing to a new perspective of statistical analysis. Eventually, the direction of arrival of the PD source can be derived. The proposed method has greatly improved the PD detection accuracy, especially in low signal to noise ratio (SNR) conditions. In simulation tests, it shows a better capability of noise immunity with signal SNR in [–5 dB, 5 dB] range. Field tests performed in a 110-kV substation (SNR around 5 dB) showed that the PD detection accuracy can be improved by up to 70% compared to the traditional methods. Our results may further promote the practical application of substation PD detection system.