Extreme Weather Prediction for Highways Based on LSTM-CNN

The weather of highway is affected by various factors, and extreme weather may cause traffic accidents and casualties. Accurate prediction of weather distribution can assist relevant personnel to make decisions, take precautions in advance and avoid disasters. The weather forecast based on the meteorological numerical model and data assimilation system does not take into account the characteristics of the highway network itself, and can not be dynamically predicted. In this paper, a weather prediction algorithm based on LSTM and CNN is used, which can accurately learn high-dimensional features from a large number of meteorological data and highway basic information, and carry out dynamic weather prediction. We use the historical weather data and road information data of a highway to verify the effectiveness of the model. Experiments show that our model has significance in the extreme weather warning of expressway and the selection and design of material type and thickness of expressway.