Study On On-line Monitoring Method Of Partial Discharge In Power Transformer

Power transformer,as an important device of the power system,is associated with safe and reliable operation of the power system.As for the power transformer in operation,the defects inside the insulation part may cause partial discharge which will accelerate the degradation of transformer insulation function and even cause electrochemical breakdown of the insulation.Therefore,it is of practical value to carry out real-time partial discharge online monitoring and diagnosis for transformer insulation.In this thesis,firstly,the in-depth research is carried out regarding to the partial discharge of transformer and a comparative analysis is conducted regarding to several partial discharge testing methods,among which ultra-high frequency testing method characterized by a strong anti-interference capacity,convenient positioning,etc.has been a research hotspot in recent years.Based on the analysis of the generation mechanism,transmission characteristic and attenuation characteristic of ultra-high electromagnetic waves under partial discharge,it is found that there is much attenuation of ultra-high frequency waves on transformer enclosure.Therefore,the ultra-high frequency electromagnetic wave sensor is placed inside transformer to improve the sensitivity of measurement.The online monitoring method of partial discharge(DSP is deemed as the monitoring and control center)is researched according to the characteristics of ultra-high frequency electromagnetic waves,the ultra-high frequency signals transmitted by the ultra-high frequency sensor are amplified and reduced,and the signals are disposed by the differential drive circuit to meet the requirement of A/D converter for analog input signals.The digital filter algorithm research is carried out and EMD wavelet packet denoising method based on the autocorrelation function is proposed with respect to the situation that partial discharge ultra-high frequency signals received include white noises.Firstly,the signals are broken into several intrinsic mode functions by virtue of EMD method,the intrinsic mode function with white noises is identified by virtue of normalized autocorrelation function.Then,the de-noising processing is conducted towards the intrinsic mode function with white noises by virtue of wavelet packet denoising algorithm.Finally,all the intrinsic mode functions are restructured to get the restructured and denoised partial discharge signals.The simulation result shows that EMD wavelet packet denoising method based on the autocorrelation function is superior to wavelet packet denoising and traditional EMD denoising algorithm and can preserve partial discharge signals better while denoising.