Research On Power Quality Disturbance Analysis Based On Segmented Improved S Transform

In recent years,with a large number of nonlinear,impact and fluctuating loads connected to the state grid,a series of power quality problems appear in the power system.They will lead to the decline of power qualtiy and further threat the stable operation of the power system.The accurate detection and identification of power quality disturbances is the basis for analysis and researching further on the causes of power quality disturbances.Which has practically important significance for ensuring the safe operation of the power system and reducing national losses.In this paper,the power quality problem is studied from three aspects,i.e,disturbance signal detection,feature extraction and classification,respectively.It describes the relevant international standards for power quality problems and researchings status.This paper first describes the relevant international standards of power quality problems and the researchings status all over the world,and then we summarizes the causes and characteristics of various disturbances.Finally summarizes the advantages and disadvantages of traditional disturbance signal detection,feature extraction and classification algorithms.Firstly,due to the poor accuracy of traditional power quality disturbance detection methods and based on the principle of S transform,a power quality disturbance detection method with improved S transform is proposed.According to the frequency of the power quality disturbance signal.The frequency range is divided into three intervals of low frequency,intermediate frequency and high frequency,and then the factor of window width is adjusted to obtain sets the window width adjustment factor to obtain better time-frequency resolution.Secondly,in view of the disadvantages of the traditional intelligent optimization algorithm that the convergence is slow and easy to fall into the local optimum,this paper overcomes the shortcomings of the original gray wolf algorithm.This article uses three common test functions to test the performance of genetic algorithm,particle swarm algorithm,original gray wolf algorithm and improved gray wolf algorithm.The simulation results show that the convergence speed and stability of the improved gray wolf algorithm been greatly improved.Finally,in view of the advantages and disadvantages of the global kernel function and the local kernel function of the support vector machine,it can improve the generalization ability of classifier by constructing a hybrid kernel function support vector machine.In order to solve the difficulty of selecting the hyperparameters of the hybrid kernel function support vector machine,we can use the improved gray wolf algorithm to optimize the hyperparameters of the hybrid kernel function support vector machine.Simulation experiments show that IGWO-HSVM can achieve higher classification accuracy under the same noise environment.