A hybrid particle swarm optimization approach for distribution network reconfiguration problem

Abstract

A hybrid particle swarm optimization (PSO) approach is proposed here to solve the distribution network reconfiguration (DNR) problem. This approach is a combination of the binary PSO algorithm and the discrete PSO algorithm. In the problem-solving process, the distribution network is simplified through grouping the branches, and then each group of branches is represented by one dimensional coding. Based on the DNR necessary condition summarized in this paper, particles are evolving regularly, and this improves the efficiency of the search process. Each cycle of iteration within the optimization process is accomplished through two steps. The first step is using the roulette bet method to choose the optimal group of branches that should be opened, and this method is based on the sigmoid ( ) function value in the binary PSO algorithm. The second step adopts the discrete PSO algorithm proposed in this paper to further select the optimal branch that should be disconnected in the group selected in the first step. Simulation on a typical 69-node case is conducted, the result of which has indicated both a rapid convergence and a good robustness of this hybrid approach.