Simulation of the breaking arc in high voltage circuit breakers based on a combination of the least square and the neural networks techniques

Abstract

Circuit breakers remain among the principal devices that are used for the control and security of electrical networks in order to ensure the interruption of the current in both normal operating or short-circuit conditions. Although appearing simple in its fundamental principle, since it consists on contacts separating, the currents cutting remains a very complicated phenomenon where several parameters intervene . At the contacts opening, these apparatuses are simultaneously submitted to various types of constraints: dielectric, thermal and electrodynamics.
In this paper, a simulation of the breaking arc in high voltage circuit breakers based on a combination of the least square and the neural networks techniques will be presented. A simulation program has been developed by the means of Matlab software that enables to adapt polynomial functions to the neural simulations. The obtained results show the efficiency of the method on a 245kV/50kA line breaker.