A distribution network with multi-type distributed generations reconfiguration model considering multiple objectives is proposed. Pareto optimal criterion and multi-objective binary particle swarm optimization (MOBPSO) algorithm are used in the reconfiguration process. The algorithm is improved in three aspects: the ring-based coding method to reduce the probability of infeasible solutions; the dynamic change of inertia factor and the asynchronous change of learning factor to improve the algorithm's adaptive ability; combination of niche sharing mechanism and proportional operator selection are proposed to update the particles. According to the decision maker’s bias, this paper finally gives an evaluation mechanism, which can select the appropriate reconstruction scheme from the optimal solution. Numerical results show that the improved MOBPSO algorithm can improve the efficiency and stability of the optimization. This paper’s mode as well as the improved algorithm cooperating with the evaluation mechanism can provide an effective reconfiguration scheme for decision maker.