Using Artificial Neural Network to estimate maximum overvoltage on cables with considering forward and backward waves

Abstract

Lightning is known to be one of the primary sources of most surges in high keraunic areas. It is well-known fact that surge overvoltage is a significant contribution in cable failures. The other source of surge voltage is due to switching and it is pronounce on extra high voltage power transmission systems. The effect of both lightning and switching surges is weakening the cable insulation. The progressive weakening of such insulation will lead to cable deterioration and eventually its failure. Each surge impulse on the cable will contribute with other factors towards cable insulation strength deterioration and ultimately cable can fail by an overvoltage level below the cable basic impulse level (BIL). The maximum lightning overvoltage for a given cable depends on a large number of parameters. This paper presents the effect of model parameters (e.g., rise time and amplitude of surge, length of cable, resistivity of the core and sheath, tower footing resistance, number of sub conductors in the phase conductor (bundle), effect of surge arrester, length of lead, relative permittivity of the insulator material outside the core, power frequency voltage, stroke location, cable joints, shunt reactors, sheath thickness) on maximum cable voltage. An Artificial Neural Network (ANN) is trained to estimate peak overvoltage generated in presence of back flashover. Levenberg-Marquardt method is used to train the multilayer perceptron neural network The simulated results presented clearly show that the proposed technique can estimate the maximum overvoltage with good accuracy.