Research On Detection Method Of Photovoltaic Grid Connected Islanding Based On WPT And A-RF

With the rapid development of industries in various countries and the increasing energy demand,photovoltaic grid connected power generation has become one of the research hotspots,but it also faces the islanding effect problem that may occur during grid-connected operation.In order to reduce the harm caused by the islanding effect to the maintenance personnel,reduce the impact on the stability of the entire power grid system,and damage the power equipment,it is of far-reaching practical significance to study efficient islanding detection methods.In order to reliably identify the islanding effect and improve the accuracy of islanding identification,it is proposed to use wavelet packet transform(WPT)to extract the islanding identification feature and combine it with the accurate random forest(A-RF)algorithm to apply it to islanding detection.By calculating the local energy entropy of each frequency band of the common coupling point(PCC)voltage signal,the wavelet packet basis function is optimized,and the wavelet packet basis function is used to decompose and reconstruct the PCC voltage signal,and calculate its characteristic energy value to form identification The input feature vector of the system.As the classification result of traditional RF algorithm is determined by the decision tree voting mode,and the classification accuracy is low,an improved algorithm for reasonable weight distribution based on the classification accuracy of the decision tree is proposed to strengthen the voting ability of the decision tree to optimize random forest identification The classification capability of the model enables accurate identification of islanding effects.In the MATLAB/Simulink simulation software,a photovoltaic grid-connected power generation system model was built.And normal,islanding and three kinds of interference operation states were set.The extracted islanding identification features were input into the model mentioned in this article and the traditional detection model for experimental analysis.The results show that compared with the traditional method,the islanding detection method based on A-RF has higher recognition accuracy and faster detection speed,and can still make accurate judgments under the interference state,which proves the feasibility and effectiveness of the detection method.The paper has 34 pictures,8 tables and 67 references.