Scientific and reasonable early warning and assessment of conductor icing can help to take accurate response measures to prevent the freezing disaster risk in time. This paper presents a multi factor grey prediction model GM (1,N) suit for small sample database. Compared with the traditional neural network model, the proposed model requires lower sample size of the modeling database and corresponds to higher modeling and calculation efficiency. The degree of conductor icing can be predicted in real time according to meteorological parameters, which can realize the risk warning of transmission line conductor icing disaster. Based on the case analysis of the proposed model, the icing degree is divided into five grades in the engineering application scenario. It is found that the average error of the multi factor prediction model based on GM (1,N) grey theory in ice thickness prediction is 8.1%, and the risk warning accuracy of transmission line icing disaster is as high as 94%. In addition, the probability of judging the high risk level as the lower one can be decreased by adding a certain safety margin value near the critical value of the ice thickness. In the ice area, The application of the ice thickness grey prediction model proposed in this paper can guide the anti?ice work of transmission lines in icing area.