基于数据全局与局部视角的风电功率异常数据清洗方法

Wind power anomaly data cleaning method based on data global and local perspectives

摘要: 针对风速-功率的异常数据难以识别清洗的问题,提出一种基于HDBSCAN-LOF的全局与局部视角的风电功率异常数据清洗方法。首先,根据风速-功率异常数据的分布特征对其进行分类；然后,分别利用HDBSCAN、LOF以及所提方法对异常数据进行清洗。最后,基于实际风场数据计算Spearman相关系数来验证所提方法的有效性。结果表明,所提方法的Spearman相关系数较HDBSCAN方法和LOF方法分别提高了0.0086和0.0829,验证了所提方法的有效性。

Abstract: Aiming at the problem that the wind speed-power abnormal data is difficult to identify and clean, a wind power abnormal data cleaning method based on HDBSCAN-LOF global and local perspectives is proposed. Firstly, it is classified according to the distribution characteristics of wind speed-power anomaly data. Then, HDBSCAN, LOF and the proposed method are used to clean the abnormal data respectively. Finally, the Spearman correlation coefficient is calculated based on the actual wind field data to verify the effectiveness of the proposed method. The results show that the Spearman correlation coefficient of the proposed method is 0.0086 and 0.0829 higher than that of the HDBSCAN method and the LOF method, respectively, which verifies the effectiveness of the proposed method.