ABSTRACT
A power quality (PQ) problem is defined in terms of deviations of waveforms of voltage and current from the sinusoidal nature and variations in the frequency. The causes and sources of PQ events are needed for initiating the PQ improvement action. Detection and classification of PQ issues will help to find the cause of these disturbances. Recently, signal processing and intelligent techniques have been employed for this purpose. An approach for the identification of single-stage PQ disturbances is proposed in this work. This technique is based on processing of the signals using the Stockwell transform to compute an output matrix in a frequency domain. A power quality index (PQI) and a PQ time location index (PQTLI) are proposed, which are computed from this matrix and used for the identification of the different types of single-stage and complex PQ issues. Six statistical features are computed from the PQI and PQTLI, which are considered as input data for the decision tree for classifying the PQ events. The Stockwell transform and rule-based decision supported technique is effective in the identification of the single-stage PQ issues. Classification accuracy of the single-stage PQ issues has been achieved as high as 98.44%. A performance comparison of the technique undertaken to compare with the accuracy of the algorithm based on the discrete wavelet transform (DWT) and artificial neural network (ANN) was found superior compared to the DWT- and ANN-based approach.
