Time-frequency based pattern recognition technique for detection and classification of power quality disturbances

This paper proposes a simple time-frequency based pattern recognition technique for detection, classification and quantification of power quality disturbance waveforms. The proposed technique consists of time-frequency analysis, feature extraction, and pattern classification. Though there are several time-frequency analysis techniques exists in the literature, this paper uses S-transform to obtain the time-frequency characteristics of power quality events because of its superior performance under noise as well as harmonics. Using the time-frequency characteristics, a set of optimal features are extracted for pattern classification of power quality disturbances. Finally, a simple rule based system is developed for detection and classification of various power quality disturbances. Although the authors have proposed recently an S-transform based fuzzy expert system for power quality detection and classification, the proposed technique is simple and 98% accurate even under the presence of harmonics and high signal to noise ratio for the most of power quality disturbances.