Combined Approach of LST-ANN for Discrimination between Transformer Inrush Current and Internal Fault

摘要

Discrimination between transformer inrush Current and an internal fault is attained by an inrush-detection algorithm to prevent the unnecessary tripping by magnetizing IC. Conventional methods of protection detection are based on comparing the 2^{nd harmonic and fundamental components of current. This paper presents a digital protection technique based on combined Least-Squares Technique (LST) and an Artificial Neural Network (ANN) to discriminate between magnetizing-inrush and internal-fault currents in three-phase power transformers. The LST is first applied to estimate the magnitudes of the fundamental, 2nd and 3rd harmonic components respectively. Second Harmonic Ratio (SHR) and third harmonic ratio (THR) are then calculated for each of the three phases. Then they fed into ANN for classifying the transient phenomenon into either IC or IF. Training and testing patterns of IC and FC over wide range of inception angles are obtained by MATLAB/Simulink simulation of a 3-phase non-linear magnetic core transformer bank.}