The invention discloses an active distribution network dynamic topology reconstitution method based on mixed artificial intelligence. Multiple improved artificial intelligence methods are adopted for optimizing solving, the superiority of various algorithms can be integrated, meanwhile the various artificial intelligence methods are improved during the process of combining various optimization algorithms, furthermore the astringency of the algorithms is improved, the defects such as locally optimal solution are avoided, three object functions are integrated in the optimization process, three different optimization objects are considered, and a corresponding Pareto optimal solution set is solved out; Since measurements of indexes such as on-off operation times and power supply volume are not constant, in order to establish a unified measurement model, all the indexes are converted into loss indexes according to respective conversion relations, and finally a dynamic analytic hierarchy process is adopted for deciding solutions in the Pareto optimal solution set, so that the optimal solution representing an optimal distribution network structure is obtained.