Aiming at the change of distribution network topology and power flow distribution caused by large-scale grid-connected distributed generation, the existing distribution network reconfiguration algorithm is not enough to deal with it. An improved Artificial Fish Swarm Algorithm(AFSA) is proposed to solve the distribution network reconfiguration with distributed generation. Aiming at the shortcomings of AFSA, such as slow convergence rate, fixed foraging direction, low flexibility, low localization and low search accuracy,using an all-round foraging behavior, and combined with differential evolution(DE) and AFSA, improves algorithm flexibility and increases population diversity, making the algorithm easy to jump out of local extremum and improve convergence accuracy. Finally, the analysis of the example shows that the proposed algorithm is effective. The results show that compared with other intelligent algorithms, the improved AFSA has better convergence precision and convergence speed, and can be applied to the reconstruction of distributed power distribution network.