The paper aims to develop an effective method to identify, detect and classify power quality disturbances using the efficient Extreme learning machine (ELM). It's important to evaluate the learning time while designing any kind of computational algorithms, which used for classification. ELM comprises of a single hidden layer Feed Forward Neural Network (SFNN) with better generalization ability and extreme fast learning. The efficient Fast S-transform(FST) is imposed to extract discriminating features of different power quality disturbances wave form and that correspondence feature will be given as input of the ELM classifier and further proceed for classification. By this process performance of FST based ELM classifier is compared with the ST based ELM classifier with distinctive features of different PQ disturbances. FST signal analysis is done by using different classifier and corresponding result is found out. Ten varieties of PQ disturbances have been chosen for the proposed classification task. The proposed FST based ELM classification is feasible and promising for a real time application as evidenced from our results.
展开▼
