Detection, Classification and Location of Short Circuit Faults in Distribution Systems Using Intelligent Systems

Abstract

This project aims to develop an intelligent system to detection, classification and location of short-circuit faults in distribution systems. There are many paper have approached artificial intelligence techniques to solving fault diagnosis problems, i.e., artificial neural networks, fuzzy logic, intelligent systems and genetic algorithms, these applications occurs because the complexity problems and inexistence of effective analytical formulations. The techniques combine the human operators' knowledge and ability to execute routines in a safe way and high speed response. These mechanisms available to operation can produce a great step forward. This project emphasizes the development of intelligent tools applied to fault diagnosis in real time, to assist operator in your decision tasks. Among the procedures present in this methodology can highlight system modeling, data acquisition, information pre-processing, extraction of features, determining network status, fault classification, fault resistance estimate and fault location. For this, it is necessary to have a technique that provides the desired reliability, robustness, flexibility and with low processing time. These requirements are the objectives of this research, which should be validated through different distributions systems simulations (small, medium and large) using computer programs, i.e., ATP (Alternative Transient Programs), Matlab and others. The project idealized model is an intelligent system also enable to generalization and continuous learning, within this perspective, may be need develop an integrated system, combining more than one intelligent artificial techniques' (in focus, artificial neural network and fuzzy logic), what is expected to produce high quality results.