Abstract: With a high proportion of distributed resources continuously connected to the distribution network, it is possible to make full use of the black start value of multiple types of distributed resources to make the distribution network system support the rapid recovery of key nodes of the main network. In order to solve this problem, a scheme of multi-type distributed resources to assist key nodes of the main network and a real-time scheduling strategy are proposed. Firstly, the advantages and feasibility of the distribution network system to support the black start of key nodes of the main network are analyzed. Then, in view of the node startup sequence and recovery path are driven, to support each other, to maximize the power distribution network system available generation capacity, minimum important load in power distribution network loss and the minimum recovery path weight of main goal to establish the key node and important load in distribution network collaborative optimization model, and combined with Floyd algorithm based on the genetic algorithm method, A real-time scheduling strategy of black start based on rolling optimization model is proposed to ensure that the power supply at key nodes of the main network can be restored safely and stably in the shortest time. Finally, a 35-node main distribution network partition system in a certain region is taken as an example to carry out simulation analysis. The results show that the proposed model and method significantly reduce the recovery time of power nodes in the main network, and effectively improve the grid resilience and distributed resource utilization efficiency.