基于风力机输出功率序列相似度匹配的风电场分群新方法

A new clustering method of wind farm based on similarity match of wind turbines output power sequence

为提高风电场等值模型的精度,需要将处于不同运行状态的风力机组进行分群,对此提出一种应用谱分析聚类算法的风电场机组分群方法。该方法以一段时间内各风力机组输出有功、无功序列间的相似度为分群指标,利用基于几何样板匹配的谱分析聚类算法对各风力机组的运行数据进行聚类分析,寻找动态运行过程中相似度较高的机组并将其划为同一个机群,最终完成对风电场内全部风力机组的分群。实际算例的仿真结果表明,等值模型与详细模型的输出特性一致,分群结果可反映该段时间内风力机组的运行特性,分群方法合理、有效。

In order to improve the accuracy of the wind farm equivalent model, it＇s necessary to cluster the wind turbines in different running states, so a clustering method of wind farm based on spectral analysis cluster algorithm is proposed. In this method, the classification of wind farm is completed by putting the wind turbines with the high similarity in the dynamic process into the same group,which takes the similarity of wind turbines output active and reactive power in a period as clustering index and adopts spectral clustering algorithm based on geometric template matching to do the clustering analysis of the wind turbine operation data. Numerical simulation results show that： equivalent model output characteristics is consistent with detailed model and clustering results can reflect the wind turbine operation characteristics, which proves that the clustering method is reasonable and effective.