风电场功率曲线异常数据的清洗与建模

风电场风速功率数据中通常包含大量异常数据,难以反映风机的真实工作情况,影响风电功率预测的准确性,进而造成一定的经济损失.针对该问题,分析异常数据的特征,提出滑差-四分位异常数据剔除方法,并利用高次多项式和Logistic函数对剔除后的数据进行风速-功率曲线建模,最后用和方差、均方根误差和确定系数验证该方法的适用性和有效性.实例分析表明,该方法简单高效、通用性强,可显著提高风电机组功率特性分析的准确度.

Cleaning and modeling of abnormal data of wind farm power curve

The wind speed and power data of wind farms usually contain a large number of abnormal data, which is difficult to reflect the real working conditions of wind turbines, affect the accuracy of wind power prediction, and then cause certain economic losses. To solve this problem, the characteristics of abnormal data are analyzed, a slip-quartile abnormal data elimination method is proposed, and the wind speed power curve of the eliminated data is modeled by using high-order polynomial and Logistic function. Finally, the applicability and effectiveness of this method are verified by variance, root means square error and determination coefficient. The analysis in the example shows that the method is simple, efficient and versatile, and can significantly improve the accuracy of wind turbine power characteristic analysis.