An Intelligent Classification Method of Series Arc Fault Models Using Deep Learning Algorithm

Abstract

In recent years, machine learning techniques have been widely used to solve many problems for fault diagnosis. A significant number of electrical connectors have been utilized in photovoltaic systems in the presence of parallel and serial modules structures, where many various faults can take place. One of these faults, known as a series arc fault that frequently happens in the PV system. Many series arc fault generator models are derived from studying this type of fault. In this paper, a new intelligent method is proposed to classify various models of series arc fault generator. Different types of series arc fault models have been simulated to generated more than 800 records. The intelligent classification method has been proposed using Python to precisely discriminate among different models structured in a way that simplifies deep feature learning, where a light convolution neural network has been used; the proposed method achieved a high accuracy 98%.