Modeling of Car-Following Behavior on Urban Underground Expressways Based on Data-Driven Methods

Abstract:

In order to reveal the operational characteristics of traffic flow on underground expressways， a car-following model based on the data-driven method was proposed by using the high-precision vehicle trajectory data obtained by driving simulator， and was calibrated and verified. First， the driving simulation experiment was conducted according to the scenario model of the east section of the North Cross Passage in Shanghai to obtain the car-following data. Next，the support vector regression （SVR） method was selected to establish the car-following model， into which the driving behavior constraints were introduced.Finally， the improved SVR-based car-following model was calibrated and verified by using the experimental data. The results show that the support-vector-regression-based car-following model considering driving behavior constraints can well describe the car-following behavior on underground expressways. The model possesses a great transplantability， which has a high accuracy on other underground expressways. The car-following model proposed in this paper can quantitatively analyze the interaction of vehicles on underground expressways， and provide the basis for traffic simulation and risk research.