Integrated driving behavior modeling for expressway merging zones based on virtual vehicles

Highway merging areas are critical traffic bottlenecks, where complex interactions between car-following and lane-changing behaviors substantially affect traffic flow efficiency and safety. Existing simulation models often treat these two driving behaviors in a decoupled manner, limiting their ability to accurately represent merging-zone dynamics. To address this limitation, this study introduces a virtual vehicle–based unified modeling framework for expressway merging areas. By mapping virtual vehicles into the target lane, the model seamlessly integrates lane-changing and car-following behaviors within a consistent decision-making structure. The framework is calibrated using real-world trajectory data via a genetic algorithm (GA) and validated through stability analysis and simulation experiments. Compared to established models such as the default Simulation of Urban MObility (SUMO), MOBIL, and Social Force (SF) models, the proposed approach demonstrates substantially better agreement with field observations across multiple performance indicators—including speed, acceleration, stop frequency, lane-changing frequency, lane-changing time-to-collision (LCTTC), and space headway. This improved simulation fidelity provides researchers and practitioners with a more reliable tool for evaluating merging-area management strategies and supports the development of more effective intelligent transportation systems for bottleneck mitigation.