Feature characterization of power quality events according to their underlying causes

This paper addresses the problem of extracting effective features for the analysis of underlying causes of power quality (PQ) disturbances. For each underlying cause, we define and extract a set of features based on analysis of voltage/current waveforms or the combination of them. The proposed feature sets are then used for building a rule-based classification framework for automatic identification of the underlying causes stored in PQ databases. These rules are based on the extracted features. Using the proposed features and rules, the proposed classifier has yielded a correct classification rate of 95.8% for a total of 96 disturbance sequences, demonstrating a high accuracy distinguishing between the different underlying causes in PQ events.