A combination of DE and SVM with feature selection for road icing forecast

Abstract

The road icing is an adverse weather condition leads to dangerous driving conditions with consequential effects on road transportation. A numerical road icing predication approach is employed for automatic prediction of road icing conditions for Shiyan City. The approach is derived from the support vector machine (SVM). To improve the classification accuracy for road icing prediction, a modified differential evolution (DE) is employed to simultaneously select features. With the data from 1980 to 2006, using the proposed approach, the road icing models for the city are created, which have been used for the prediction for Shiyan City from 2007 to 2008. The results have shown feasibility and effectiveness of the forecast approach.