Clasificación y ubicación de fallas en líneas de transmisión utilizando el algoritmo de retro-propagación del clasificador

Abstract

The research developed is based on designing and training an artificial neural network (ANN) to locate and classify faults in electrical transmission systems (SET), since they are built to cover long stretches, and therefore are exposed to adverse conditions which cause electrical failures. Therefore, the research paper proposes the design of a neural network in Matlab - Simulink software, using the back propagation algorithm. A static study was carried out to obtain data using PowerFactory software and to consider in the neural network, after the simulation the data of short-circuit current and short-circuit angle were obtained, about two hundred and fifty faults were generated in the nine-bar system to generate a matrix of binary numbers, so that when they are entered they perform the action of locating and classifying. A comparison was made with the compressed census method with an efficiency of 96% and where the neural network method has reached an efficiency of 95%, which shows that the use of neural networks is currently a viable method, which through constant training achieves greater efficiency and reliability of the ANN model.