A novel methodology for power transformer differential protection by incorporating artificial neural network

The core task of the differential relay is to protect the power transformer under faulty condition and to exercise the restrain function in other operating conditions. This paper proposes a better protection scheme by incorporating ANN along with differential relay. The pattern classification property of ANN is exploited which distinguishes the various operating conditions occurring in the power transformer and gives the trip command when internal fault occurs. This scheme is superlative as it is least affected by the establishment of harmonics in the differential current. The model of this ANN based scheme comprises of one ANN structure using FFBP learning method. The learning data are obtained by simulation of a power system network having a transformer using MATLAB. The regression plot and performance plot clearly indicate that the model is fast and more accurate in recognizing and classifying the catastrophic condition occurring in the power transformer.