Least-square B-spline approximation based wind turbine power curve modeling

摘要

The power curve of a wind turbine plays a significant role in design, operation, control and monitoring of wind turbines. Especially, it provides necessary information for wind farm maintenance decisions, allowing for not only comparison among the same type of wind turbines installed in different place but also the same one at different time stamps. In this study, a standardized wind turbine power curve modeling procedure is proposed based on real supervisory control and data acquisition (SCADA) data in order to obtain an accurate power curve for performance evaluation in an online manner with the smoothness guaranteed. The data are firstly preprocessed for anomaly detection by following normalization correction. The whole range of available wind speed and power are both segmented into slices in modeling and the centroids are computed after outlier detection. The least-square B-spline approximation method is adopted for power curve construction based on dominant points selected from the centroids. Finally, the test results based on actual operating SCADA data demonstrate better performance in comparison with two counterparts.