Intelligent Fault Detection and Diagnosis System Design for Wind Turbine Based on FAST

The dynamics of a wind turbine model are simulated in five different scenarios, i.e., normal operating conditions, short circuit in generator, wear and Delamination in gearbox and low-speed-shaft and aerodynamic asymmetry. In this paper, first of all, Principal Component Analysis is used to extract features from input data. Secondly, two methods are employed to isolate faults of different types at different locations: Support vector machines and Multi-layer perceptron. The proposed approach are validated on a 1.5-MW threeblade wind turbine using the National Renewable Energy Laboratory wind turbine simulator FAST (Fatigue, Aerodynamics, Structures, and Turbulence) code.