The prediction and diagnosis of wind turbine faults

The rapid expansion of wind farms has drawn attention to operations and maintenance issues. Condition monitoring solutions have been developed to detect and diagnose abnormalities of various wind turbine subsystems with the goal of reducing operations and maintenance costs. This paper explores fault data provided by the supervisory control and data acquisition system and offers fault prediction at three levels: (1) fault and no-fault prediction; (2) fault category (severity); and (3) the specific fault prediction. For each level, the emerging faults are predicted 5–60 min before they occur. Various data-mining algorithms have been applied to develop models predicting possible faults. Computational results validating the models are provided. The research limitations are discussed.