A large number of photovoltaic(PV) systems are decentralized and connected to distribution networks, leading to challenges in model simulation. This paper presents an equivalent modeling method for regional decentralized PV clusters based on cluster analysis. The proposed method is based on the grouping principle that PV systems have similar dynamic response characteristics, and grouping results are obtained using a fuzzy C-mean (FCM) clustering algorithm. Based on the clustering evaluation index, the optimal grouping number can then be determined and PV systems in the same group are combined into an equivalent PV system. The method to acquire equivalent parameters for PV arrays, transformers, and lines is also presented. A multi-machine equivalent model of regional decentralized PV clusters is established, and the precision and efficiency of the proposed method is demonstrated using an actual distribution network.