Aiming at the problem of distribution network reconfiguration with wind-solar-load uncertainty, a distributed generation and load output model was established. A multi-objective function was constructed with system operating cost and voltage offset. Firstly, an improved particle swarm optimization and K-means (IPSO-Kmeans) clustering algorithm was proposed to divide the typical daily load curve. Then, the improved Harris Hawk optimization algorithm (IHHO) was applied to the distribution network reconfiguration for optimization calculation. In order to improve the Harris hawk optimization (HHO) algorithm, the population distribution was uneven, the optimal solution space range cannot be completely searched, and it was easy to fall into local convergence. The good point set was introduced to generate population initialization to improve the uniformity of population search space. The explorer position update formula in the sparrow search algorithm was combined with the position update formula in the HHO algorithm exploration stage. This improved the global search ability of the algorithm. The Cauchy-Gaussian mutation perturbation strategy was used to jump out of the local optimal solution. Finally, the simulation results of IEEE33 node distribution network system show the effectiveness of the proposed method.