A Data Mining Based Protection and Classification of Transients for Two-Core Symmetric Phase Angle Regulators

This paper investigates the applicability of time and time-frequency features based classifiers to distinguish internal faults other transients - magnetizing inrush, sympathetic external with current transformer saturation, overexcitation for Indirect Symmetrical Phase Angle Regulating Transformers (ISPAR). Then faulty unit (series/exciting) ISPAR is located, or else transient disturbance identified. An event detector detects variation in differential currents registers one-cycle 3-phase post samples which are used extract training seven classifiers. Three different sets wavelet coefficients, time-domain features, combination energy obtained from exhaustive search using Decision Tree, random forest feature selection, maximum Relevance Minimum Redundancy used. The fault detected a balanced accuracy 99.9%, localized 98.7% no-fault classified 99.5%. results show potential accurate detection localization, identification. proposed scheme can supervise operation existing microprocessor-based relays resulting higher stability dependability. modeled simulated PSCAD/EMTDC by varying several parameters.