Icing forecast theoretical models in existence consider so many factors and have very complicated structures, but they still have errors in forecasting. Meanwhile, icing forecast models based on regression methods think less about time cumulative effect, thus the forecasting results are still different with the real conditions. On the basis of the review of nowadays icing forecast models, a transmission line icing forecast model considering the time cumulative effect of meteorological factors is proposed in this paper. Based on the built icing thickness growth model, the relationship of icing thickness and time is analyzed. According to the analysis result and the influence of various meteorological factors on icing growth, the degree of icing growth under different meteorological factors is given. The degree can be calculated by historical meteorological data and together with historical icing thickness data, it can be used to train SVM regression forecasting model to forecast icing thickness. The proposed model is validated through a case analysis and demonstrated to have the forecasting precise advantage over typical icing forecast models in existence. This work is supported by National Natural Science Foundation of China (No. 51707018), China Postdoctoral Science Foundation (No. 2017M612907), and State Grid Xinjiang Electric Power Co., Ltd. (No. 5230DK16001P).