Research Of Power Quality Disturbances Detection And Recognition Algorithms

Nowadays,with the rapid development of science and technology,various new power technology and electric devices emerge in an endless stream,on the other hand,the power demand of the whole society is expanding rapidly.This makes a variety of problems related to power quality increasing,and threatens the security of the power grid and relevant electrical equipment,causing huge losses to the social economy.Detection and identification of power quality disturbance signal is the basis of solving power quality related problems.Only by correctly detecting and identifying disturbance signal can power quality problems be further processed and analyzed.In this paper,eight kinds of common power quality disturbance signals are established,and power quality disturbance is analyzed in three aspects: disturbance preprocessing,detection and recognition.In the aspect of disturbance signal pretreatment,facing the actual situation that noise often exists in power quality disturbance signal in power system,and noise will affect the detection and recognition of power quality disturbance signal.Based on the analysis and understanding of the decomposition and reconstruction of wavelet transform,an improved wavelet threshold functions denoising algorithm is proposed in this paper.The simulation results show that the improved threshold functions denoising algorithm can effectively remove the noise in power quality disturbance signal.At the same time,compared with the traditional hard threshold functions and soft threshold functions denoising algorithm,the improved wavelet threshold functions denoising algorithm has better denoising effect.In the aspect of disturbance signal detection,this paper focuses on the S-transform and its modulus matrix.Based on the theory of S-transform,the S-transform modulus matrix is obtained by S-transform of various disturbance signals.From the S-transform matrix,a variety of characteristic variables,such as mutation point curve,timeamplitude squared sum-mean curve,are extracted,and used for disturbance start-stop time,amplitude and frequency of disturbance signal.Detection of the Etc.At the same time,because of Heisenberg uncertainty principle,it is impossible to simultaneously improve the analytical force in both time and domain.A scheme to improve the adjustment factor of S-transform Gauss window is proposed,and the feasibility of the scheme is verified by simulation.In the aspect of disturbance signal recognition,in order to reduce the problem that disturbance signal can not be accurately expressed when the pretreatment of disturbance signal and the error of artificial feature selection are made,a new statistical feature,waveform amplitude statistical feature,is proposed in this paper.It is used in the classification of power quality disturbance signal.The amplitude of disturbance signal is divided into 40 intervals,and the sampling point number of each interval is counted.The one-dimensional vector characteristic composed of the number of sample points is the statistical characteristic of waveform amplitude.At the same time,the disturbance signal classifier is obtained by training the features with the artificial neural network,which realizes the recognition and classification of power quality disturbance signals.In order to simulate the actual situation in reality,this paper also adds noise to the eight kinds of power quality disturbance signals.The simulation results show that the neural network recognition and classification algorithm based on waveform amplitude characteristics can effectively classify power quality disturbance signals to a certain extent.At the same time,compared with the traditional db4 wavelet features,the statistical characteristics of waveform amplitude have a stronger ability to express power quality disturbance signals.