Under the "carbon peaking and carbon neutrality goals", the power grid faces intensive upgrades in form and function. The characteristics of nodalization and modularization are more prominent. It is urgent to determine a routing algorithm taking into account the characteristics of nodes to ensure the supporting capacity of the communication network. First, this paper defines importance degree based on the nodes’ power grid influence degree and topology relationship. Then the importance degree is embedded into the policy model from the state transition and adaptive search convergence control, so as to guide the route optimization direction. Then a parameter knowledge model is introduced to mine the relevant knowledge of the problem and cooperate with the importance degree to guide the optimization. This reduces the parameter sensitivity of the ant colony algorithm. Finally, the simulation results show that, in needing to meet the quality of service (QoS), the knowledge-based routing policy considering the importance of nodes can converge quickly, and provide favorable and stable performance in primary and standby routing.