Classification Method Of Power Quality Disturbances Based On CBR And SVM-KNN

With the development of electrical engineering and industrial automation,power quality has become a modern enterprise can not be ignored factors. Any power quality problems will lead to a decline in the quality of products,and even project pause caused immeasurable losses to the user. The development of information technology on power quality and power supply reliability put forward higher requirements, At the same time,in the modern power system,the power electronic equipment is used more and more widely, various nonlinear, shock, volatility load for an increase in the number,makes power system suffered more serious quality pollution. In order to improve the quality of the user’s power supply, to solve various problems caused by power quality,realize the harmonious and sustainable development of the national economy. As the first step in the electric energy quality of governance, correct classification and research of various power quality phenomena appeared, indepth analysis to explore the electromagnetic interference problem is yet to be known, and the electric energy quality disturbance problems to achieve effective classification and to eventually solve the power quality problems is necessary and is the basis of research and analysis of power quality problems.This thesis presents a new approach based on Case-Based Reasoning(CBR) and Support Vector Machine(SVM) for accurate classification of power quality disturbances signals. Firstly, original power quality disturbance signals are processed by atomic decomposition, features were extracted, the results are used to build the case base of power quality signals. Then the SVM-KNN approach is used for case retrival in the classification, find the most similar original cases. At last, to determine the disturbance signal classification results by modifying or reusing the most similar case results. Numerical results show that this classifier combine the respective advantages of CBR and SVM. The method has less training samples, more fast classification speed and higher classification accuracy percentage. Six types of power quality disturbances are accurate identified. The simulation verifies its validity to classify power quality disturbance.