Application of artificial intelligence methods for determination of transients in the power system

Abstract

Numerous factors, including sudden load reductions, switching transient loads, lightning strikes, and malfunctions of control devices, can result in overvoltage. Overvoltage can harm associated power supply components and result in insulation failure, electronic component damage, flashovers, etc. A machine learning technique called a "neural network" estimates computation results that depend on a lot of inputs. For a variety of reasons, neural networks have recently been used to manage and optimize the power system. This paper presents an artificial neural network (ANN)-based approach to determining overvoltages in power systems. To simulate overvoltages, many simulations were performed in Electromagnetic Transient Program (EMTP). Variations of parameters of interest that have an influence on overvoltages were made using JavaScript that was connected to EMTP models. The extraction of characteristic parameters from overvoltage waveshape is a demanding task, and it was conducted in MATLAB, as was a overvoltage classification methodology based on neural networks. Results were presented and discussed.