Research On High Voltage Partial Discharge Data Processing System

The operation state of high-quality insulation of electrical equipment is the prerequisite for safe and stable power grid.At present,most of power equipment faults come from the insulation failure.Paying attention to the deterioration of the insulation in electrical equipment is critical for the staff of the power sector.The partial discharge acts as an indicator of insulation integrity degeneration and differential partial discharge signals represent different types of insulation deterioration.Furthermore,insulation performance evaluation of electrical equipment can be achieved through data analysis of partial discharge.By judging corresponding information of the partial discharge signal,it is better to determine the maintenance time and avoid blind maintenance.The denoising and the identification of partial discharge are not only the way to realize the insulation performance judgment of electrical equipment,but also the core of partial discharge data processing system.This paper analyzes the causes,types and related characteristics of local partial discharge signals in the switchgear and partial discharge signals in the cable.On this basis,the partial discharge pattern recognition should be studied,which includes denoising,feature extraction and pattern recognition of partial discharge signals.For the interference,this paper shows the theory of wavelet transform and multi-resolution analysis.With analyzing the hard threshold function and soft threshold function,an improved threshold function is aimed to optimize the denoising effect.Through simulation verification,the SNR of filtered signal is higher.The next step is the partial discharge feature extraction,the method of combining multi-resolution analysis and the wavelet packet energy is adopted.And,the detailed steps of its implementation are introduced.The characteristics obtained by applying the above method lays the foundation for the next intelligent algorithm pattern recognition.Then the article shows basic theory of artificial neural network and BP neural network is selected to establish pattern recognizer.The designed parameters are analyzed.Finally,the combined feature quantities are input to the BP recognizer for simulation and verification.The result shows that this method can achieve partial discharge signal classification with high accuracy.Based on the above theory and simulation,a partial discharge data processing system is established to realize the functions of automatic classification of partial discharge signals,occupancy of various types of partial discharge signals,and specific display of single discharge pulse waveforms.