The communication system is the carrier of grid and information network coupling. The construction of reliable Smart Grid and Ubiquitous Power Internet of Things depends on the high coupling between the power grid and the information network. With the development of” the Grid and Internet”, the risk assessment of the power grid communication system has the characteristics of various levels, complex indexes and strong correlation. To deal with the risk effectively, this paper proposes an index system for evaluating the operation risk of grid communication system, which includes the cable loss level, the communication network power supply risk, the equipment operation risk, the external environment risk, the business interruption risk and the management risk. At the same time, a risk assessment model is built based on a fuzzy neural network, which overcomes the problem that traditional risk assessment methods rely too much on expert experience and makes prior knowledge embedded in the network. Finally, the model is proved to be scientific and practical by an example analysis.