Evaluation of lightning performance of transmission lines protected by metal oxide surge arresters using artificial intelligence techniques

Abstract

Lightning and switching overvoltages are the main causes for faults in electrical networks. In the last decades, several different conventional methodologies have been used for the adjustment of the lightning performance of high voltage transmission lines, which are protected against lightning using overhead ground wires and surge arresters. The current paper proposes a new developed Artificial Neural Network (ANN), based on the Q-learning algorithm, in order to estimate the lightning failure rate of lines of the Hellenic system. The results obtained by the ANN model exhibit a satisfactory correlation in comparison with the real recorded data or the simulations results taken from a conventional method.