New technique for categorization of Power Quality disturbances

Providing effective classification techniques for various power quality (PQ) events is gaining the attention of the research community. The process of power quality analysis and diagnosis is a complex one for many reasons, including the complex modeling of power systems, the extensive amount of system data that is currently available through PQ monitors, and the lack of expert knowledge. Therefore, it is evident that computerized system analysis is vital for the realization of effective and efficient power quality diagnosis systems. In this paper two intelligent techniques are developed that perform power quality classification functions. These techniques are based on wavelet analysis, subtractive cluster algorithms and Artificial Neural Networks (ANN). Many signals are generated to simulate different types of power quality phenomena then wavelet analysis is applied to these signals. Different feature extraction methods are proposed to reduce the amount of processed data which dramatically improves the performance of the proposed PQ classifier compared to other techniques proposed elsewhere. The extracted features are then used to train different ANNs.