Diseño de un sistema de detección de fallos en el generador de una turbina eólica basado en inteligencia artificial

Abstract

CONDITION FOR PUBLICATION OF PROJECT. The present project aims to develop a model of the Doubly-Fed Induction Generator (DFIG) used in wind energy systems, to represent its behavior during electrical faults and propose an early fault detection system. To this end, a coupled simulation model was established through the integration of OpenFAST and Simulink, enabling a detailed representation of both the wind turbine subsystems and the electrical grid, and reproducing the wind system dynamics with high fidelity. A mathematical model was implemented to simulate inter-turn short-circuit (ITSC) faults in the stator winding was implemented, generating fault scenarios with different levels of severity. Based on this model, representative operational data were generated for both healthy and faulty conditions of the generator. Using these data, an unsupervised artificial intelligence–based fault detection system was developed, capable of identifying anomalies almost immediately after their occurrence, achieving average detection times of less than one second. The results demonstrate that the proposed model accurately captures the transient response of the DFIG under electrical faults and serves as an effective tool for the early detection of electrical faults in wind energy systems. Keywords: Doubly Fed Induction Generator, Interturn short-circuit fault, Unsupervised learning, Early fault detection.