| Road alignment is the important influencing factor of vehicle safety traffic, the design quality of road alignment decides the road safe condition. At present, the top-priority road alignment design method of our country is the method of the design speed. The limitation of this method is design road alignment by a stationary design speed. But in fact, the vehicle speed in the course of driving may not fixed. The disadvantages coming from this design method is that national highway traffic accidents rate and accident mortality is far above the western developed countries.The road alignment design method of the vehicle speed is is studied extensively at home and abroad, it's based on the realistic vehicle speed, fully considering the vehicle driving laws, pilot needs and road traffic conditions, therefore, this design method more close to reality, can effectively improve road safety situation, has important significance to national transportation business.The key of this design method is vehicle speed prediction models, this article analyse the variation regularity of the vehicle speed on highway from the perspective of road alignment. without considering the driver condition, vehicle differences and mutual interference, the external expression of drivers estimate to road alignment is the selection of vehicle speed. Therefore, the study of speed prediction model, is not only the key of road alignment design also has important reference value for road alignment assessment.Firstly ,this paper introduces the research status of vehicle speed and road alignment, then, proposes the vehicle speed prediction methods based on road alignment, analyzes the relevance between vehicle speed and road alignment, gradually get the variation regularity of the vehicle speed. Based on the research results, build the vehicle speed prediction models by the method of linear regression. The validity of the prediction models is proved by statistical tests and actual measuring datas. Final, some research results is obtained by the application research of the vehicle speed prediction models. |