Safety Evaluation Model of Vehicle Behaviors on Two lane Highway

Abstract:

A safety evaluation model was established based on neural networks, fuzzy evaluation method and space vector theory, for an objective evaluation of vehicle behaviors safety on two lane highway. First, according to the previous research, overspeed index, the average headway and safety overtaking sight distance were selected as discriminant indexes of safety state about the three types: overspeed behavior of free flow, following behavior and passing behavior， respectively, and safety level were determined for each indicator. Then, the relationship model between the seven major indicators of the operating environment on two lane highway and vehicle behaviors characteristics was established by three layer neural network model and its reliability was verified with the measured sections data. The membership function about the discrimination indexes of safety state was established by fuzzy evaluation method and the safety level of the three types of vehicle behaviors were divided. Vehicle behaviors comprehensive safety was characterized by comprehensive evaluation vector which was proposed by space vector theory. The evaluation model is illustrated by a case study which will serve a reliable method to evaluate vehicle behaviors safety on two lane highway.