Investigation of Effective Automatic Recognition Systems of Power-Quality Events

There is a need to analyze power-quality (PQ) signals and to extract their distinctive features to take preventative actions in power systems. This paper offers an effective solution to automatically classify PQ signals using Hilbert and Clarke Transforms as new feature extraction techniques. Both techniques accommodate Nearest Neighbor Technique for automatic recognition of PQ events. The Hilbert transform is introduced as single-phase monitoring technique, while with the Clarke Transformation all the three-phases can be monitored simultaneously. The performance of each technique is compared with the most recent techniques (S-Transform and Wavelet Transform) using an extensive number of simulated PQ events that are divided into nine classes. In addition, the paper investigates the optimum selection of number of neighbors to minimize the classification errors in Nearest Neighbor Technique.