Power grid information system working in a continuous working environment, the operational state of the risk frequency is high, a network information system based on information clustering fusion processing risk warning technology for the operational state of the grid information system. The information extraction model of grid information system running state risk assessment is constructed, and the collected data of grid information system operation state evaluation is analyzed by statistical clustering. Fuzzy adaptive C clustering algorithm is used to evaluate the running state and classify the state attributes of the power network information system. The spectrum parameters reflecting the risk characteristics of the network information system running state are extracted. The simulation results show that this method has high reliability for early warning of power grid information system operation state, high accuracy and low false alarm rate, and real-time accuracy, so as to ensure the stable operation of power grid information system.