In order to improve the assistant decision-making for rush-repair scheduling for distribution network on the power grid emergency management system, an optimization model for multiple failures repair in distribution network is established, which has comprehensively considered the repair resource allocation, multi-group collaboration and rush-repair order. In addition, the multiple populations coevolution mechanism is introduced to enhance the optimization capability of artificial bee colony algorithm for high dimensional complex problem. A unified scheduling method combined with the modified artificial bee colony algorithm is designed. Simulation results of a PG&E69-bus system show that the proposed method can rapidly provide pre-arranged emergency rush-repair scheme after the occurrence of multiple failures in distribution network to minimize the economic loss due to outage. This work is supported by Science and Technology Project of State Grid Corporation of China (No. SGSDDK 00KJJS1700535).