Traditional strategy for multi-fault repair of distribution relies on decision-maker to make the subjective judgment,which is apparently lack of scientific basis. It is easy to make mistakes in judgment. The emergency repair resources cannot be reasonably allocated and the power supply cannot be restored at the first time. In order to solve the problem,an optimization model for multi-fault repair in distribution system is established,which has comprehensively considered multi-group collaboration and rush-repair order. The non dominated sorting genetic algorithm (NSGA)-Ⅱ algorithm with adaptive parameters is designed. After the Pareto front is obtained,the knee solution algorithm based on angle selection can directly solve a relatively ideal scheme without the participation of decision maker. Simulation results of the distribution network in a town show that the adaptive parameter adjustment strategy can improve the global search ability in the early stage of population evolution and the local search ability in the later stage of evolution. The final repairing scheme obtained by the knee solution algorithm based on angle selection can be selected directly from multiple feasible schemes,which can reduce the burden of decision-makers and it is suitable for practical emergency repair work.