Clustering of Photovoltaic Power Stations Based on the Improved AP Clustering Algorithm

Yawen Ding,Fei Gao,Yingwei Jiang,Kai Hou,Donghui Zhang,Xuewei Shi

DOI: https://doi.org/10.1109/iecon51785.2023.10312567

2023-01-01

Abstract:To deal with the problem that uncertain parameter in Affinity Propagation (AP) clustering algorithm affects the clustering results, this paper proposes a clustering method based on AP clustering algorithm and Genetic Algorithm (GA) for Photovoltaic (PV) power stations. Firstly, the modeling of PV power stations is established and clustering index are determined according to the dynamic characteristics of PV power stations. Then the proposed method takes the silhouette index, which is used to evaluate the quality of the clustering result, as the fitness function of GA. The bias parameter placed on the main diagonal of the similarity matrix in the AP clustering algorithm is constantly iterated and updated until obtaining the optimization result. Finally, the studied case verifies the better clustering performance of the proposed method. The 20 PV power stations in the Taoling Substation can be simplified into three groups, reducing the complexity and calculation time for equivalent modeling.