Distribution network reconfiguration is an important measure for optimal operation of distribution network. Because of its clean and renewable distributed generation, it is more and more introduced to distribution network reconfiguration. In view of the output random DG's influence on the distribution network reconfiguration, the scene division method based on wasserstein distance index is introduced into the distribution network reconfiguration and the wind power is divided into a number of scenes. Considering the randomness of wind power and load fluctuations, this paper puts forward the optimal scheme to determine strategy based on binary particle swarm optimization algorithm and establishes the model of distribution network reconfiguration. A loop coding and initial population generation strategy is put forward based on the rule “out the same and reserve the different”, the computation efficiency is improved. The established model is solved in the IEEE33 node distribution network system. The feasibility and effectiveness of the proposed strategy are verified and an optimal reconstruction scheme is obtained, which is adapted to the randomness of wind power and the load fluctuation.