A novel approach using a FIRANN for fault detection and direction estimation for high-voltage transmission lines

This paper presents a novel approach to fault detection, faulted phase selection, and direction estimation based on artificial neural networks (ANNs). The suggested approach uses the finite impulse response artificial neural network (FIRANN) with the same structure and parameters in each relaying location. Our main objective in this work is to find a fast relay design with a detection time not dependent on fault conditions (i.e., current transformer saturation, dynamic arcing faults, short-circuit level, and system topology) and that uses only unfiltered voltage and current samples at 2 kHz. The suggested relay, which we have named FIRANN-DSDST, is composed of a FIRANN together with post-processing elements. The FIRANN is trained globally using training patterns from more than one relaying position in order to be as general as possible. The FIRANN is trained using an improved training algorithm, which depends on a new synaptic weights updating method, which we have named the mixed updating technique. The proposed relay is trained using training patterns created by simulating a real 400-kV network from the Spanish transmission network (R.E.E.). Finally, the proposed relay is tested using simulated and real fault data. The results encourage the use of this technology in a protective relaying field.