An accepted paper shared early by Scientific Reports on 13 October compares support vector machines (SVM) and an artificial neural network (ANN) with a Random Forest baseline for detecting wind-turbine faults. Its abstract describes simulated operating data generated in FAST and MATLAB/Simulink under turbulent wind conditions. Monitored variables include generator speed, generated power and blade-pitch angles; evaluation uses five-fold blocked temporal cross-validation.

The abstract reports that performance depends strongly on fault type, with ANN benefiting some challenging patterns. This does not establish one superior model for every fault. The accessible publication contains the early abstract and author information; complete methods and result tables could not be verified in this review. These simulation findings are not presented as evidence of a system validated on operating turbines or a commercial deployment.