Abstract—Cable faults is an important issue for underground cable lines, and the sheath current is a major factor for cable faults. The sheath current occurs on metallic sheath of high voltage underground cable. Thus, metallic sheath of underground cable is grounded to prevent the sheath current faults. If the sheath current which will generate on metallic sheath, the most suitable grounding method can be selected to prevent the sheath current effects. In this study, artificial neural network (ANN) is used to determine the sheath current of underground cable line, and many simulation studies of the sheath current are made to obtain training data of ANN. Simulation studies of sheath current of different underground cable lines are made in PSCAD/EMTDC. Also optimization methods are used to update weights of ANN. Gravitational search algorithm (GSA) and differential evolution algorithm (DEA) are used to optimize weights of ANN. It is seen that training and forecasting errors of hybrid methods are lower than classic ANN.

Index Terms—Artificial neural network, differential evolution algorithm, gravitational search algorithm, underground cable.

Bahadir Akbal is with Department of Electrical and Electronic Engineering, Selçuk University, Konya, Turkey.

Cite:Bahadir Akbal, "GSA-ANN and DEA-ANN Methods to Prevent Underground Cable Line Faults," International Journal of Computer and Electrical Engineering vol. 10, no. 2, pp. 85-93, 2018.