Power quality event classification using complex wavelets phasor models and customized convolution neural network

Origin and triggers of power quality (PQ) events must be identified in prior, order to take preventive steps enhance quality. However it is important identify, localize classify the PQ determine causes origins disturbances. In this paper a novel algorithm presented voltage variations into six different considering space phasor model (SPM) diagrams, dual tree complex wavelet transforms (DTCWT) sub bands convolution neural network (CNN) model. The input data converted SPM data, transformed using 2D DTCWT low pass high which are simultaneously processed by CNN perform classification events. proposed method based on Google Net trained with default configuration as deep designer MATLAB environment. achieve higher accuracy reduced training time than compared reported event methods.