A new statistical-based algorithm for CT saturation detection utilizing residual-based similarity index

A statistical based method for current transformer saturation detection is presented in this paper. This method which simply detects and determines the saturation point starting and ending in real-time, is based on new formulation of correlation-based feature selection. This method has immunity against noisy condition. Based on similarity of saturated and unsaturated output currents, proposed index so-called saturation index (SI) is calculated. Unsaturated output current is estimated utilizing least squares (LS) technique. Both Simulation and implementation are conducted in MATLAB software environment. The investigation have been carried out taking different condition such as presence of remnant flux, high-resistance burden and presence of decaying DC component into account. The results demonstrate that the proposed method has pretty good performance under different circumstances. In addition it only utilizes 10 samples (equal to 1/20 cycle) make a decision and as a result it is suitable for real-time application.