High-speed directional relaying using adaptive neuro-fuzzy inference system and fundamental component of currents

This paper proposes an adaptive neuro-fuzzy approach for fault direction estimation in sectional transmission lines. The ANFIS (adaptive neuro-fuzzy inference system) network is designed by selecting different input and output member functions and rules for training and testing of fault cases. The fundamental component of current obtained from three-phase current employing discrete Fourier transform (DFT) is given as input to the ANFIS module. The trained ANFIS module is then tested for detecting the fault direction. The relay is located at middle section-2, which is considered as the primary section to be protected. It takes section-1 as reverse section and section-3 as forward section. This method is not affected by the variation of fault type, fault inception angle, fault location, and fault resistance. The biggest advantage of the ANFIS method is that it can detect the fault direction within 1 ms in almost all cases, which is much less than the implemented distance relaying scheme. The second advantage of the method is that it takes less number of training samples to detect the direction accurately as compared to other training algorithms like ANN, SVM, etc. The third advantage of the proposed scheme is that it offers protection to 99% of line length in all the three sections. (c) 2015 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.