中文摘要：

为解决多维风光联合发电相关性建模问题,提出结合K-means聚类和藤结构原理建立混合藤Copula模型。考虑光伏出力昼夜周期性,利用混合藤Copula模型重点分析风光联合出力在日间的相关性,并在该模型的基础上结合回溯搜索算法对电力系统进行无功优化。以美国某地区相邻2个风电场、1个光伏电场的实测数据为例,在IEEE 30节点系统中对所提方法进行验证。算例结果表明,所提方法能够更准确地描述多维风光出力的相关性,并且利用该方法建立的无功优化模型能有效降低网损,减少节点电压偏差和发电机无功偏差。

英文摘要：

To solve problem of describing relevance among outputs of multiple wind and PV farms, this article proposed a mixture model combining K-means clustering method and vine copula structure, considering PV circadian periodicity and using mixture vine copula structure to analyze correlation of daytime data. Backtracking search algorithm is utilized based on this model to accomplish reactive power optimization. The proposed method was verified with simulation in IEEE 30 system with data measured from two wind farms and a PV farm in the United States. Simulation results indicate that the proposed method could turn out accurate description of correlation among multiple wind and PV farms, and get more reliable results of probabilistic load flow.