Power Quality Disturbances Detection and Classification Based on Deep Convolution Auto-Encoder Networks

Power quality issues are required to be addressed properly in forthcoming era of smart meters, grids and increase renewable energy integration. In this paper, Deep Auto-encoder (DAE) networks is proposed for power disturbance (PQD) classification its location detection without using complex signal processing techniques classifiers. technique, Gabor filter used extract a set general features from the convolution PQD image. Subsequently, through sparse based DAE network, essential optimal extracted learnt which by simple classifier (SoftMax) classify type. Furthermore, temporal information obtained image correctly locate disturbance’s initiating terminating instants. The network has benefits Learning-based terms automatic feature selection, but it requires smaller data sets. issue obtaining optimised, robust, strong thus resolved. Excellent accuracy with appropriate parameter setting network. technique compared three other methods i.e. support vector machine (SVM), stacked auto-encoder (SAE) principal component analysis (PCA) implementing all four on python platform same set. It an overall more than 97% at noise ratio (SNR) 20dB, higher side when under noisy environment. Additionally, method less computation time alternative approaches like SVM. Thus, outperforms existing single multi-disturbance greater reduced complexity time.