Abstract: Transmission line icing trigger wire off shares,tower breakage and other accidents,a serious threat to the safe operation of the power grid,so accurately predict the transmission line icing conditions is important for the safe and stable operation of the power system.A comprehensive identification method based on the icing process of micro-meteorological transmission lines is proposed.The method includes polynomial regression model,time series analysis model and machine learning model,using Matlab system identification toolbox and libsvm toolbox is more convenient and comprehensive on icing identification and modeling.Firstly,the researchers process the acquired data,and then select the appropriate model structure for parameter identification to obtain the initial model.Finally,the initial model is verified and the final identification model is obtained.Experiments show that the final identification model can predict the icing process of transmission lines and accurately reflect the actual ice weight of transmission lines,and improve the convenience and comprehensiveness of the model.