Abstract: Because of the voltage over-limit problem in the scenario of large-scale distributed photovoltaic (DPV) access to the distribution network, the clustering method adopted to regulate voltage could easily cause uneven distribution tasks. This paper proposes a voltage regulation strategy for the DPV cluster that considers payload balancing. Firstly, the electrical distance between distribution network nodes is obtained through the payload concept, which is used to construct the Modularity Index. Secondly, an Equilibrium Index is established based on load balancing theory, which combines the Modularity Index and the Davies-Bouldin Index to construct a Comprehensive Index to improve the spectral clustering algorithm for partitioning the distribution network into clusters and selecting the key nodes. Then, the minimum voltage deviation and network loss are selected as the objective function to establish the reactive power voltage control model, and particle swarm optimization (PSO) is used to address this. Finally, simulation verification is conducted on an IEEE 33-node system under DPV access. The results show that compared to the comparative index and algorithms, the cluster partition scale and cluster capacity deviation are relatively optimal in this paper. Moreover, the computation time is saved, and node voltage deviation as the network loss is smaller during the maximum DPV output period. The distribution of voltage regulation tasks and payload allocation are more reasonable, confirming the proposed method's effectiveness.