PQ EVENT DETECTION AND CLASSIFICATION BASED ON DUAL TREE COMPLEX WAVELET AND COARSER-FINE TWO STAGE ANN CLASSIFIER

摘要

Dual Tree Complex Wavelet Transform (DTCWT) is shift invariant and has 2m redundancy as compared with Discrete Wavelet Transform. In this paper, DTCWT is used to obtain non-redundant sub bands energy levels representing PQ events in different and unique sub bands. The sub bands are quantized and thresholded to retain 95% of information that will improve classification process. Two stage FFNN architecture is designed to classify six possible PQ events by performing coarse and fine classification process. The two stage classifier with 10 neurons in each FFNN architecture and 4 neurons in the second stage achieves 97.5% of classification accuracy. The develop algorithm is suitable for real time applications in smart meters.