Cluster Partitioning Method for Distribution Networks with Distributed Photovoltaics Considering Voltage Stability

Abstract

To address the voltage stability degradation and increased topological complexity caused by the growing penetration of distributed photovoltaic systems in distribution networks, this paper proposes a voltage-stability-oriented cluster partitioning method. First, a comprehensive cluster partitioning index is established, considering both structural and functional aspects of the distribution network framework. Structurally, modularity metrics are used to quantify network aggregation characteristics, while functionally, dynamic evaluations integrate voltage stability indices and source-load matching indicators. Second, to overcome premature convergence in existing clustering algorithms, an improved coati optimization algorithm is proposed. This algorithm employs chaotic mapping for uniform population distribution, introduces adaptive escape operators to balance global exploration and local exploitation and incorporates optical opposition-based learning to enhance its ability to escape local optima. Finally, simulations on the IEEE 33-bus system demonstrate that the proposed comprehensive clustering index improves voltage stability by 46.79%, 49.54% and 47.83% compared to single indices, respectively. The improved clustering algorithm reduces computation time by 62.79% and 65.22% compared to the traditional coati algorithm and moth-flame optimization algorithm, while achieving 97.72% of centralized voltage control effectiveness with an 88.84% faster response time, verifying the effectiveness of the proposed method.