Power Quality Events classification using Customized Convolution Neural Network

Integration of renewable energy resources in power grid may provide technical and economic benefits but their potential impact on the voltage, current quality in grid takes a major role. The integration of RES to power grid results in power quality disturbances (PQD). Identifying these PQD’s in real time has become considerably more challenging. Due to electronic system maintenance, PQDs may cause security difficulties and loss. To minimizing PQ events, the events must be identified and classified so that appropriate preventative action may be taken. This paper focus on most commonly occurring voltage quantity events, which have severe impact on equipment connected to the grid in comparison with all other power quality events. In this paper a novel technique has been used for classifying voltage changes into six distinct PQ occurrences. The voltage data is transformed to SPM data, which has separated into sub bands using 2D DTCWT and are analysed by the 2D CNN model at the same time to categorize PQ events. When compared to previously published PQ event classification algorithms, the suggested approach achieves improved accuracy in event classification while maintaining a simpler architecture.