Research Of De-noise And Pattern-recognition In Partial Discharge Detection System

Power transformer is a kind of most important instruments in power system, and its safety running will directly influence the credibility of power supply. Partial discharge is the main reason for insulation ageing of high-voltage electrical equipment, and also an important symptom and manifestation of insulation ageing. So, for keeping power transformer normal, it is crucial to monitor the partial discharge signal and determine the types of partial discharge effectively and precisely.This thesis discusses on the partial discharge phenomenon in power transformers. For the problem of denoising of the partial discharge signal,this paper introduces a new method, Wavelet threshold filtering based on Emprirical Mode Decompositon. What's more, the author has a study on the method to eliminate the noise of the signal collected from power transformer. The results from simulation and measured data indicate that using this algorithm a batter interference suppression can be obtained.This thesis has discussed the methods of the fingerprint charts of the PD parameter extracting,and introduces the BP neural network to the partial discharge pattern recognition.Through to carrying on the modeling,sample and is trained and tests actually because of the three layer perceiving machine of BP nenural network,and then identify to three types of partial discharge.At the end, this thesis develops a software system of transformer partial discharge based on LABVIEW and MATLAB.This system can realize several functions such as eliminating noise from power transformer partial discharge sigal,displaying the spectrum of the partial discharge signal,recognitizing the type of the partial discharge signal and so on.As proved and experimented, the software system designed in this thesis can effectively denoise partial discharge signal and determine the type of partial discharge. The system can be applied to partial discharge on-line monitoring system...