Prediction of Wind Turbine Performance Parameters Based on Virtual Models
DescriptionThe purpose of this project is to develop models for predicting three critical performance parameters, generated power, vibration at drive-train and tower, of a target wind turbine based on data collected from other wind turbines. Statistical methods, data mining algorithms and ensemble algorithms, will be applied for extracting parametric and non-parametric models from collected data. A comparative analysis will be performed to evaluate capabilities of proposed approaches in developing accurate prediction models. Models able to offer accurate prediction results can be considered as virtual sensors for measuring wind turbine performance. Virtual sensors can be applied to assist the fault identification and condition monitoring of wind turbines to reduce the wind farm maintenance cost. Wind turbine operational data collected from a commercial wind farm in Iowa, USA, and the wind power station in Lamma Island, Hong Kong, are utilized in this research.
|Effective start/end date||1/10/12 → 20/05/15|