Time-Frequency Representation of Resistance for Modeling of Transformer Winding Under Impulse Test

Abstract

In power engineering, transformer fault detection during impulse tests has always been an important topic. Accurate diagnosis of faults can significantly enhance the safety, reliability, and economics of the power system. Modern computerized fault detection techniques require information on current and voltage waveforms of transformer windings subjected to impulse voltages under different fault conditions. Simulation studies based on digital models of transformer windings greatly enhance the understanding of surge voltage and current waveforms. The conventional lumped parameter transformer model for digital simulations does not consider frequency-dependent parameters accurately and extensively. The nonstationary nature of voltage and current waveforms inside the transformer winding requires the model parameter to be both time and frequency dependent. This paper proposes a wavelet transform-based time-varying frequency-dependent winding resistance model for transformers under impulse test. The model has been validated by comparing real-life and simulated current responses of transformers over a wide range of power ratings. Digital modeling of transformers using the proposed time-frequency-dependent parameters will help in constructing accurate models of transformers windings for simulation studies.