Neural Network Based Classification Research On Power Quality Disturbances In Hohhot

As the rapid development and popularization of computer technology, storage technology and Internet in the modern society, the amount of system data in electric power industry is becoming increasingly large at an exponential rate. How to make full use of the data has become one of the key issues in the power industry, as well as analyzing, processing and refining these data rapidly and effectively. In this paper, data mining theories and technologies are employed to solve the classification problems of electric energy quality. Taking the electric energy quality disturbance data in Hohhot area as examples, the cases of electric energy quality disturbance are analyzed in the paper, so as to ensure the rational quality management of the electrical energy and improve the overall power grid operation level.In this paper, according to phase space reconstruction theory, a novel feature extraction method of electric energy quality is proposed to fulfill the feature extraction tasks of electric energy quality disturbance cases. Based on support vector machine and self-organizing neural network technologies, the classification model of electric energy quality disturbance is proposed. Moreover, analysis and comparisons on Hohhot typical residential power quality data are performed, the results show that the SOM based classification method holds a higher classification accuracy.