Design of a Generic Fault Diagnosis Model for Electrical Distribution Networks Using a Support Vector Machine (SVM) Algorithm

Abstract

This article presents the design and implementation of a generic model for fault diagnosis in electrical distribution networks, based on the Support Vector Machine (SVM) algorithm. The proposal arises in response to the increasing complexity of modern distribution systems, which significantly heightens the need for rapid and accurate fault detection and localization. In this context, a computational model was developed within a Python programming environment and its performance was evaluated using the IEEE 34-bus standard test feeder. To facilitate the analysis, the network was segmented into seven zones, across which four common fault types observed in distribution systems were simulated: single line-to-ground fault, line-to-line fault, double line-to-ground fault, and three-phase fault. Simulations were performed using the specialized software DIgSILENT PowerFactory, generating a dataset representative of fault-related behavior. This dataset was subsequently preprocessed and balanced to ensure an adequate class distribution and then used in the training and validation phases of the SVM classifier. The results obtained from the simulations confirm that the proposed model achieves high levels of accuracy in both fault type identification and fault zone localization. These findings support the conclusion that the SVM-based approach constitutes a robust and effective tool for enhancing fault detection and diagnosis processes in real-world electrical distribution systems, thereby contributing to a safer and more reliable system operation.