Research on the weaving area capacity of freeways under man–machine mixed traffic flow

With the maturity and development of autonomous driving technology, mixed autonomous and manual vehicle traffic will become the main form of traffic flow in the future. First, conservative and radical lane change models of manual and autonomous vehicles in the weaving area are constructed to further study the freeway traffic capacity. Additionally, this study uses US-101 and Jinbao interchange trajectory data to fit the conservative-radical model switching point distribution function for manual vehicles. Moreover, simulation experiments determine the optimal switching point position of an autonomous vehicle lane-changing model to optimize traffic flow in the weaving area. Based on gap acceptance theory, the method considers the distribution of model switching points, estimates the ideal safe gap for each type of lane change, calculates the occurrence probability of different lane change types of vehicles, and determines the expected number of lane-changing vehicles in each weaving lane. Then, a capacity model of the weaving area considering man–machine mixed traffic flow is constructed using linear programming. Then, this study uses MATLAB to determine the optimal value of the linear programming model and compares the SUMO simulation results with the theoretical analysis results. The results show that when the autonomous vehicle penetration rate is 20%, 50% and 80%, the deviation rates of the optimal solution of the theoretical model and the simulation results are 4.7%, 9.1%, and 9.2%, respectively, verifying the reliability of the traffic capacity model. The proposed traffic capacity model of the weaving area under mixed traffic flow can provide technical support for the planning and design of the weaving area and improving its traffic operation state. •A capacity model includes safety gaps and probabilities for each lane change type.•Considered difference of lane change behaviours between MVs and AVs in safety gap.•Considered switching of lane change model of AV and MV in the probability.•Conducted a SUMO simulation experiment to validate the capacity model's efficacy.