Abstract: In order to meet the reliable transmission requirement of massive service data in the power grid, this paper proposes an adaptive risk equalization method for power services considering dual-network coupling characteristics. First, a dual-network coupling model of power grid and power communication backbone network is constructed, within which the importance of power services is evaluated considering the influence of grid nodes. Furthermore, the risks faced by service transmission are analyzed from the service layer, transmission layer and physical layer, and the risk equalization optimization problem based on adaptive weight is established. Finally, a learning-based auction algorithm is used to adapt to the dynamics of service arrival and network state, solve the spectrum competition conflict, and realize the adaptive risk equalization. Also, theoretical analysis of the performance loss with unknown global information is given. The simulation results show that the proposed method can effectively improve the reliable bearing capacity of power communication backbone network, and reduce the service transmission risk compared with traditional methods.