基于线路阻抗聚类的分布式光伏电站等效建模

EQUIVALENT MODELING OF DISTRIBUTED PHOTOVOLTAIC POWER STATIONS BASED ON LINE IMPEDANCE CLUSTERING

为对大规模分布式光伏电站进行建模以研究其运行特性,提出一种分布式电站聚类等效建模方法。根据电站分布式结构特点,该方法将各光伏发电单元至公共连接点的等效线路阻抗列入聚类算法的聚类指标中,将各光伏发电单元聚类并将同一类中的光伏发电单元进行聚合以构建多机等效电路模型。其中,针对传统的模糊C均值聚类算法(FCM)因过度依赖初始聚类中心点而易陷入局部最优的缺点,该方法使用Canopy-FCM聚类算法进行改进,在FCM算法前用Canopy聚类算法进行预处理。最后通过RT-LAB实时仿真平台对所提聚类等效建模方法的合理性和精确性进行验证。

In order to model a large-scale distributed photovoltaic power plant and analyze its operation characteristics,a clustering equivalent modeling method for distributed power plants is proposed in this paper.In this method,according to the distributed structure characteristics of the power plant,the equivalent line impedance of each photovoltaic power generation unit to the point of common coupling(PCC)is included as an extra index of the clustering.The clustering algorithm is then used to cluster each photovoltaic power generation unit into different classes and the multi-class photovoltaic units are aggregated to build a multi-machine equivalent circuits model.Canopy clustering algorithm was first used to pre-process the data before using the fuzzy C-means algorithm for clustering,so as to improve the disadvantage that the over-dependence on the initial clustering center may lead to the fuzzy C-means algorithm falling into local optimum.Finally,experiments based on RT-LAB real-time simulation platform are carried out to verify the rationality and accuracy of the proposed clustering equivalent model.