When power system operates under abnormal conditions, a large number of alarms will be received by EMS. A fast, correct and comprehensive alarm-processing system is of great importance to help the operators make necessary decisions. Equipment fault is the essential course of power system disturbance and fault. Based on this, this paper brings up a new smart alarm method. First, equipment status evaluation is done online. Thus, fault hypothesis set is obtained by calculating equipment history operation risk and judging health status. Then, all the hypothesis is checked by multi-level flow model (MFM) which has been built before. The final answer comes out after comparing similarities between the hypothesis and the fact. Simulation results of the standard IEEE 9-bus system shows that the proposed method is doing well in both accuracy and speed and has bright future prospect in the alarm-processing field.