Power Quality Disturbance Identification Based On Multi-Feature Extraction And Optimization Of Neural Network

Power quality disturbances are complex and diverse.Rapid detection and accurate classification of various types of disturbances is the premise and basis for improving power quality.In this paper,a power quality disturbance recognition method based on multi feature extraction and optimization neural network is researched.Multiple types of feature vectors are extracted,and the best feature combination is evaluated and screened,and the classification of power quality disturbances is realized.Firstly,several typical power quality disturbance signals are described,in view of the limitation of single extraction method,the problem of comprehensive feature information can not be extracted.Wavelet Transform(WT)and Hilbert Huang Transform(HHT)are used to extract 12 eigenvalues,including energy value,energy entropy,standard deviation,kurtosis,mean and amplitude.Through the simulation analysis,the separation space of the extracted feature vectors is presented visually in 3d view,and the validity of the extracted classification eigenvector is verified.Secondly,there may be many problems with irrelevant and redundant information for the high dimension of extracted features,the multiple types of feature extraction are evaluated,and the redundancy vectors are eliminated after quantifying the diversity of each characteristic quantity.A hybrid feature selection method is proposed to rank the residual features,and select the best feature subset according to the classification algorithm,which is used for subsequent classification and recognition.Thirdly,for traditional BP classifier,the convergence time is long and slow,and the global optimum is not guaranteed.On the basis of population optimization,a new efficient algorithm combining improved Particle Swarm Optimization(PSO)and Differential Evolution(DE)algorithm is proposed,and the BP network is optimized by using the improved PSO-DE algorithm.Finally,using the optimized BP neural network,a variety power quality disturbance signals are classified.Through comparison and analysis,,and combined with several sets of actual signals in IE database,the classification effect of the method is verified.The results show that the algorithm proposed in this paper has high precision,good stability and strong anti-interference ability.