SVM based characterization and classification of composite power quality events using HHT

Power quality (PQ) issues vary from high speed events such as voltage impulses and transients, high frequency noise to wave shape faults, voltage swells and sags and total power loss. Emerging power system requires a comprehensive knowledge of power quality issues. Thus, it is important to characterize the PQ events to know the origin and underlying cause of disturbances so that appropriate preventive and corrective measures can be taken. This paper examines the effectiveness of Hilbert Huang transform (HHT) to detect and localize different composite PQ events and demonstrates classification of composite PQ events utilizing HHT based extracted features by support vector machine (SVM) classifier. Six different composite PQ events are considered in this study. The results show that all six composite events, harmonics with sag, harmonics with swell, capacitor switching with harmonics, capacitor switching with sag and capacitor switching with swell were effectively classified by SVM classifier using feature extracted by HHT.