Aplicações de redes neurais artificiais na detecção e identificação de faltas monofásicas em sistemas de distribuição
Loading...
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universidade Federal de Juiz de Fora (UFJF)
DOI
Abstract
In this work methods to detect and identify defects in radial distribution systems are developed using artificial neural networks. Several types of single-fase defects were studied, such as: high impedance defects, open-conductor defects and simultaneous defects. First of all, each of these types of defects were analysed separately in a trial system The objective was to determine the best ways to detect them. Computational simulations were carried out with the communication between Matlab and OpenDSS and they were divided into two categories. The first category uses currents information from meters allocated in the system as input data for neural networks to detect and identify each one of the types of defects treated in this work. In simulations of this category, the main objective was to analyze the impact of the number of meters and the type of measurement on the neural networks results. When more than one meter was allocated, the system was divided into detection areas. This division may have a higher cost of implementation and even computational processing, but it facilitates the process of defects location because it reduces the search area, in other words, the area where the defect may have occurred. The second category of computer simulations uses voltage information of terminal buses as input data for neural networks for detection of opening or simultaneous defects in predefined areas. The detection of defects occurred separately for each phase of the system, that is, each neural network created detected defects in a certain area of a given phase of the system. The computational results of both categories were good, showing the robustness of the structures of neural networks developed. In this work, real-time simulations were also carried out to more realistc tests of the neural networks. For this,
dSPACE device was used to model neural networks and the RTDS was used to simulate the distribution system. The results obtained were also good, showing the pratical application and the robustness of the employed technique.
Description
Keywords
Faltas de alta impedância. , Defeitos de abertura de condutores , Defeitos simultâneos , Sistemas de distribuição de energia , Redes neurais artificiais , Simulações em tempo real , High impedance faults , Open conductor defects , Energy distribution systems , Real time simulations , Artificial neural networks
Citation
Endorsement
Review
Supplemented By
Referenced By
Creative Commons license
Except where otherwise noted, this item's license is described as Acesso Aberto

