Diseño de un modelo genérico de diagnóstico de fallas en redes eléctricas de distribución utilizando el algoritmo Support Vector Machine (SVM)

Abstract

In this paper, a diagnostic model based on Support Vector Machines (SVM) was developed for the detection and identification of faults in electrical distribution networks. This research arose from the growing complexity of electrical distribution networks, which increases the challenges for rapid and accurate fault detection. As a result of this problem, there is a need to design an intelligent model capable of identifying and locating faults in electrical distribution systems.

The study was applied in a Python programming environment and on an IEEE 34-bus test network, which was classified across seven zones, considering single-phase to ground, two-phase to ground, two-phase, and three-phase faults. A database was generated from the simulations performed in PowerFactory. Subsequently, the data obtained was organized, processed, and balanced for training and evaluation in the SVM model.

The results demonstrate high performance in fault identification and location, achieving 98% accuracy across the entire system, which shows that the SVM model is an effective tool for detecting faults within real electrical distribution systems.