Ice-covering phenomenon is easy to appear on the surface of wind turbine blades during the cold wave, which will lead to problems such as reduced power generation capacity, unstable equipment operation and even failure. Therefore, it is of great significance to carry out the research on the early warning method of wind turbine ice-covering. This paper analyzes the SCADA operation database, constructs the feature quantity based on wind speed, power and temperature data, and establishes an early warning model for the occurrence of ice-covering events by using the random forest algorithm; through the real-time monitoring of the thickness of the ice cover by the rotating cylindrical array device, it establishes a real-time early warning model for the occurrence of the ice-covering events and real-time early warning dynamic mechanism. With the case of ice cover of 3.2MW wind turbine in Chongqing Wanbao wind farm, we carry out the test verification of ice cover warning. The results show that the test results of the ice-covering event occurrence warning model have a classification accuracy rate of more than 95%, and the ice-covering event warning is issued several times within 1 hour before the ice-covering situation of the wind turbine blade; the real-time warning model continues to issue warnings after the wind turbine is covered with ice, which shows that the model is able to continue to track the trend of the wind turbine"s ice-covering environment; and it is verified that the dynamic warning model can provide a decision-making basis for the safe operation and effective management of the wind turbine. It is verified that the dynamic warning model can provide a decision-making basis for the safe operation and effective management of wind turbines.