Diagnosis of PQ Disturbances using Local mean decomposition based SVD entropy and modified K-means clustering

This paper presents a novel approach for diagnosing and classifying frequently encountered islanding and power quality (PQ) disturbances in a multiple distributed generation (DG) system. To obtain noise-decomposed signals, the Local mode decomposition (LMD) technique is initially applied that results in a series of product functions (PFs). The selection of the sensitive PF component consisting of the most sensitive information related to the type of fault is determined by the singular value decomposition (SVD) based entropy method. Further to extract the underlying physical characteristics from the selected PF components, two feature extraction indices (FEIs) i.e. energy operator (EO) and mutual information entropy (MIE) are proposed resulting in a feature matrix. Finally, the obtained feature matrix is normalized and given as input to the modified K-means clustering method for classifying the proposed PQDs. With maximum classification accuracy (CA) and less computational time, the proposed system proves its efficacy and robustness in comparison to the traditional K-means clustering method.