Reconstruction of distribution network is a complex nonlinear combinatorial optimization problem. This paper presents an improved binary quantum particle swarm optimization (BQPSO), in order to overcome the basic optimization algorithm is easy to fall into local optimal solution. The algorithm is used to solve the reconstruction of the distribution network model with distributed generation (DG). By introducing the crossover operation and mutation operation of genetic algorithm to jump out precocity to improve the global search ability of the algorithm, the performance of the algorithm is improved. And the appropriate way of infeasible solution treatment is selected to improve the computational efficiency of the algorithm. Finally, through the IEEE33 node distribution system simulation, it is verified that the algorithm proposed is better in solving the reconstruction problem, and convergence speed and global optimization ability are improved.