Abstract

Transformer Differential Protection (TDP) algorithms are essential for the reliable detection of internal transformer faults by instantaneously comparing primary and secondary phase currents. Under normal operating conditions, the differential current is expected to be negligible; however, magnetizing inrush currents during transformer energization can produce significant transient differential currents, making it challenging to distinguish between inrush phenomena and genuine internal faults. To address this challenge, this study proposes a Selective Frequency-Based Transformer Differential Protection (SF-TDP) scheme designed to enhance transformer selectivity by remaining inactive during inrush transients while preserving fast and reliable response to internal faults. The proposed scheme integrates a Dual-Tree Complex Wavelet Transform (DT-CWT) energy and difference functions for feature extraction, providing improved sensitivity for early inter-turn fault detection. The traditional 87T TDP method is accordingly updated and validated through comprehensive MATLAB/Simulink electromagnetic transient simulations, demonstrating 99.4% inrush discrimination, zero false trips, average fault detection times of 5.7 ms, and successful detection of incipient inter-turn faults involving as few as 2% of winding turns, which conventional schemes failed to detect

Key words: Dual-tree complex wavelet transform; Inter-turn fault detection; Magnetizing inrush currents; Transformer differential protection