DETECTIONOF POWER QUALITY DISTURBANCES

Abstract

Automated and quick fault detection has received quite a lot of importance and some comprehensive studies have been done because of interlinking of varieties of disturbances in the power system. It takes ideal sinusoidal signal as training data aiming to recognize the other different types of faults, it generally involves two problems, i.e., selection and matching between the training and the testing data. Many studies have either studied the two independently or only focusing on selection part with less focus on the matching part of the algorithm. In this paper we propose the algorithm of transfer subspace learning to address the problem of matching which is of considerable importance as how good be the selection if the matching to particular fault is not accurate it will not give desired results. In the experiment we calculate the projection matrix and maximum mean discrepancy matrix to identify the type of fault which has occurred. The experiment so performed on the industrial data verifies our experiment to be workable in the real world situations