An improved particle swarm optimization(PSO)algorithm is presented and applied to reactive power partitioning of power grid. The method is based on complicated network theory and adopts modularity as the standard to quantitatively evaluate the quality of power network partitioning. Furthermore

a power system network model weighted by electrical distance and combined with improved PSO algorithm for reactive power network partitioning is proposed. A new particle coding method and location updating method are adopted by the improved PSO algorithm

which improves the convergence rate of heuristic algorithm aimed at the modularity and saves storage capacity. The reactive power network obtained by the improved PSO algorithm has strong regional decoupling characteristics of close electrical link internally but sparse externally. The rationality and effectiveness of the algorithm are verified by the application of IEEE 39-bus system

IEEE 118-bus system and large power network systems.