Main content area

Wind turbine performance degradation assessment based on a novel similarity metric for machine performance curves

Jia, Xiaodong, Jin, Chao, Buzza, Matt, Wang, Wei, Lee, Jay
Renewable energy 2016 v.99 pp. 1191-1201
algorithms, assets, data collection, prediction, principal component analysis, renewable energy sources, wind, wind turbines
Prognostics and Health Management (PHM) can offer substantial improvements in reliability and availability of the wind turbine asset. Driven by reducing the Operation and Maintenance (O&M) cost of wind turbines, many research efforts have been conducted to realize reliable wind turbine performance degradation assessment. Despite these efforts, it is still challenging to assess the actual degradation trend of wind turbine which will be suitable for prediction analysis. In this study, a novel similarity metric for machine performance curves is proposed and a framework of wind turbine performance assessment methodology is presented. The proposed algorithm evaluates the health condition of wind turbine by performing principal component analysis on the quasi-linear region of the power curve. The proposed methodology has been validated on a dataset collected from a large scale onshore wind turbine for a period of two years. The result exhibits a gradual degradation trend of wind turbine and indicates the ability of proposed approach to trend and assess the turbine degradation before downtime happens. The result from the proposed method also reveals its robustness to wind resolution in the power curve, which still exhibits a very similar degradation trend when the wind resolution of power curve has been down sampled.