COMPARATIVE ANALYSIS OF MLP-RBF BASED NETWORKS FOR DETECTION AND CLASSIFICATION OF POWER QUALITY DISTURBANCES

Electrical energy is first and foremost criterion for overall economic growth of the society. Widespread expansion of automated electronic equipments and miniaturization in micro-electronics used in power system hampered the quality of power a lot. This paper deals with the comparative analysis of Multilayer Perceptron Neural Network and Radial Basis Function based classifier for the detection and classification of power quality disturbances. Simple statistical parameters are used as input feature space for detailed design optimization of MLP-NN and RBF classifier. Further, for the dimensionality reduction, Principal component analysis and Sensitivity analysis are also examined. Optimized classifier is robust enough to classify the fundamental Power Quality disturbances with classification accuracy up to 99.81%.