The wildfire is one of the main causes leading transmission line tripping，which endangers the operation of power grid seriously. In order to improve the wildfire prevention level of overhead transmission lines，a fire risk level prediction model of overhead transmission lines is proposed. This model is established based on the weather index of forest fire model as well as the impact of surface environment and historical fire behavior. Then a graph-based optimization theory is introduced to determine the weight of indexes. Finally，the forecast precipitation data of 3 km×3 km resolution is used to modify the risk level of the wildfire. This model has been successfully applied to Wildfire Monitoring and Early Warning Center of China Southern Power Grid. Based on the risk prediction，the transmission operation and maintenance staffs can inspect the high fire-risk transmission lines in advance. It reduces the probability of the overhead transmission line tripping caused by wildfires，which improves the safety and stability performance of the power grid.