The Study Of Detection And Classification Of The Power Quality Disturbance Based On Hybrid Dictionary

The importance of electricity to modern society is self-evident.With the access to a large number of precision instruments and power electronic equipment,people's daily production and life also put forward higher requirements for power quality.The quality of power affects the power grid.Economic and safe operation,so for the accurate and efficient detection of power quality disturbance signals,it is necessary to study countermeasures to eliminate the disturbance signals.This paper mainly focuses on the detection,feature extraction and classification algorithms of power quality disturbance signals.The simulation of MATLAB and the experiments of actual power quality data are used to verify the effectiveness of the algorithm.Aiming at the detection and classification of interference signals in power quality,this paper proposes a method of sparse decomposition of signals by using over-complete mixed dictionary,which has a good effect on both single and composite disturbance signals.The algorithm firstly need structure including discrete sine pulse dictionary,a dictionary and discrete cosine dictionary,a dictionary,and then use the hybrid dictionary for all kinds of power quality signal sparse decomposition,detail signal and approximate signal is obtained,and extract features of power quality disturbance to single and composite signal detection and classification of power quality disturbance.A generalized orthogonal matching tracking algorithm is proposed to reconstruct the power quality disturbance signal,aiming at the problem of poor efficiency and poor recovery precision of searching the optimal time-frequency atoms in the matching tracking algorithm.On the basis of matching time-frequency atom framework,the algorithm increases the number of time-frequency atoms selected for each time to optimize the calculation,improves the calculation efficiency and reduces the error of sparse reconstruction of signals.In this paper,a classification and recognition model is built based on the random forest algorithm.Finally,the corresponding classifier is obtained through training.Through simulation,it is verified that this method can realize the classification of signals with relatively high accuracy.Finally,this paper further verified the algorithm based on 8 kinds of power quality transient disturbance signals collected from 110 kV substation of lingshou station,including 2 kinds of compound disturbance.Numerical results showed that the algorithm could still achieve effective recognition effect.