Digital protection of DFIG-based wind farms using machine learning and FRT‑based timing adjustment

With the increasing penetration of wind power plants based on doubly fed induction generators (DFIGs) in modern power systems, the development of accurate, fast, and reliable protection schemes that ensure compliance with fault ride-through (FRT) requirements has become essential. A key challenge in this domain arises from the inherently nonlinear dynamic behavior of DFIG-based wind turbines and the influence of low-voltage ride-through (LVRT) strategies on fault-induced voltage and current components, which can adversely affect protection performance by causing relay maloperation or undesired delays in tripping. This paper presents a comprehensive and intelligent protection framework based on machine learning for a radially configured wind farm, encompassing all protection zones, including turbine branches, collector lines, intermediate feeders, the step-up transformer, and the high-voltage transmission line. In the proposed method, a two-layer neural network is employed to detect faults and determine relay protection roles (primary or backup), while an enhanced FRT-based adaptive analytical algorithm is developed to adjust relay operating times. Voltage and current signals are analyzed through multiresolution analysis (MRA) based on the discrete wavelet transform (DWT), enabling the extraction of the most significant features for subsequent processing. Simulation results obtained in the MATLAB/Simulink environment indicate that the proposed protection framework significantly improves both the speed and accuracy of fault detection and decision-making, while ensuring full compliance with fault ride-through (FRT) requirements, and delivering stable, reliable, and resilient performance under dynamic network conditions.