Abstract: After a large-scale and long-term blackout accident in the distribution network under extreme disaster events, the active distribution network with distributed energy resource and emergency power supply vehicles（EPS） can form islands to supply power for the load. Before the completion of fault repair, aiming at the problem that the unfairness of the sustainable restoration time of each key load with the same weight leads to the decrease of customer satisfaction. This paper establishes a two-layer model of dynamic islanding balanced recovery, which considering the scheduling of EPS. It sets the optimal balance of critical load restoration and the maximum economic benefit of the distribution network as the upper and lower objective functions, respectively,and realizes the fair and balanced restoration of loads with the same weight in the outage period. At the same time, in order to reduce the cost of EPS scheduling and load outage loss, based on the above two-layer model and Dijkstra algorithm, an optimal scheduling method of EPS supported by the traffic network model is established, and the best candidate access node and optimal scheduling path are obtained. Then, the original mathematical model is transformed into a mixed-integer linear programming problem by using Big-M and other linearization methods. Finally, a modified PG＆E 69-bus model and a traffic network are used to verify the effectiveness of the proposed method for improving the reliability, equilibrium and economy of power supply in the fault stage.