Decision Tree based discrimination between inrush currents and internal faults in power transformer

Abstract

This paper presents a new approach for discriminating inrush currents from internal faults in case of power transformer protection using Decision Tree (DT). Differential current of one cycle inrush and internal faults are taken as inputs to the Decision Tree with corresponding target outputs as ‘0’ for inrush and ‘1’ for internal fault conditions. The DT is trained with inrush currents and internal faults with wide variations in operating parameters of the power network and provides accurate results in distinguishing inrush currents from internal faults. The proposed DT based algorithm is found to be accurate and robust, and thus highly suitable for power transformer protection.