风电场弃风异常数据簇的特征及处理方法

发布时间：2018-06-16 12:50

本文选题：风电功率曲线 + 弃风　；参考：《电力系统自动化》2014年21期

【摘要】：风电场的历史运行数据尤其是风速和风电功率数据对风电场的运行管理和电力系统的运行调度都具有重要意义。在实际运行中,风电场的弃风现象较为严重,弃风会导致风速—功率散点图中存在大量横向分布的堆积型异常数据,这会对构造风电场的等值功率曲线产生较大的影响,从而降低风电功率预测精度,进而对风电场的运行管理和电力系统的运行调度造成不利影响。文中在分析风电场弃风异常数据特征的基础上,提出一种基于四分位法和聚类分析的异常数据组合筛选模型,首先采用两次四分位法剔除常规的分散型异常数据,再使用聚类法剔除堆积型异常数据,并采用二次聚类的思想来解决k-means聚类中k的取值问题。算例分析表明,该模型可有效剔除弃风造成的异常数据簇,对不同的风电机组和风电场有较强的通用性,具有一定的工程实用价值。
[Abstract]:The historical data of the wind farm, especially the wind speed and the wind power data, is of great significance to the operation and management of the wind farm and the operation and dispatch of the power system. In the actual operation, the wind field has a serious discarding wind phenomenon. It has a great influence on the equivalent power curve of the structure of the wind farm, thus reducing the prediction accuracy of wind power, and causing adverse effects on the operation management of the wind farm and the operation and dispatching of the power system. On the basis of the analysis of the abnormal data characteristics of the wind farm abandoned wind, an anomaly number based on the four division method and the cluster analysis is proposed. According to the combined screening model, the two times four division method is used to eliminate the conventional dispersed abnormal data, and then the clustering method is used to eliminate the accumulation of abnormal data, and the idea of two clustering is used to solve the problem of K value in K-means clustering. The unit and the wind farm have strong versatility and have certain engineering practical value.
【作者单位】：中国农业大学信息与电气工程学院;
【基金】：国家自然科学基金资助项目(51174290,51477174) 高等学校博士学科点专项科研基金(博导类)资助项目(20110008110042)~~
【分类号】：TM614

【参考文献】