Aiming at the node scheduling problem of electric power communication network, this paper proposes a resource intelligent scheduling method based on federated reinforcement learning for electric power communication network to realize collaborative planning and flexible resource scheduling of provincial and local electric power communication methods. The node scheduling method in this paper is based on the tasks issued by the task nodes, the power communication network nodes according to the network attributes, power characteristics, computing attributes and data characteristics to form the power communication network node fitness, in the power communication network node fitness evaluation based on the selection of the order set of top-ranked work nodes, the formation of the work layer topology, to complete the construction of the federal learning. The working layer topology obtained using the power communication network node orchestration method proposed in this paper significantly reduces the cost consumption in the federated learning mode, helps to realize the grid-communication system model fusion and data coherence, and improves the intelligent level of the power grid-communication system resource management and network service planning.