Optimal Radial Basis Function Neural Network power transformer differential protection

This paper presents a new algorithm for protection of power transformer by using optimal radial basis function neural network (ORBFNN). ORBFNN based technique is applied by amalgamating the conventional differential protection scheme of power transformer and internal faults are precisely discriminated from inrush condition. The proposed method neither depend on any threshold nor the presence of harmonic contain in differential current. The RBFNN is designed by using particle swarm optimization (PSO) technique. The proposed RBFNN model has faster learning and detecting capability than the conventional neural networks. A comparison in the performance of the proposed ORBFNN and more commonly reported feed forward back propagation neural network (FFBPNN), in literature, is made. The simulations of different faults, over-excitation, and switching conditions on three different power transformers are performed by using PSCAD/EMTDC software and presented algorithm is evaluated by using MATLAB. The test results show that the new algorithm is quick and accurate.