Assessing Localized Contamination Severity on SiR Insulator Using Mathematical Morphology Based mRMR Aided SVM

This paper proposes a methodology of Mathematical Morphology (MM) based mRMR (Maximum Relevance Minimum Redundancy) aided Support Vector Machine (SVM) to sense the localization of contamination severity (LoCS) of 11 kV Silicone Rubber (SiR) insulator. For this purpose, several insulator samples are prepared having different localized contaminated severity. Leakage current (LC) signals are recorded across the prepared test insulators to investigate the proposed model. The statistical features are extracted from the recorded LC signals. From those MM features, highly significant features are selected using mRMR technique. Finally, the Support Vector Machine (SVM) with quadratic polynomial kernel function are developed to sense the localization of the contamination severity of 11 kV SiR insulator. The estimated results shows that the proposed technique suitably sense the LoCS of the 11 kV SiR insulator.