Micro-traffic flow modeling is essential for simulating diverse driving behaviors and developing virtual traffic scenarios for intelligent driving platforms. Integrating driver style characteristics into these simulations enhances the realism of road condition representations, providing critical environmental insights. This paper proposes a personalized micro-traffic flow model that accounts for driver styles. Driving behavior data is collected via a real-vehicle platform and processed using principal component analysis and K-Means++ clustering to extract and classify characteristic parameters of driving styles. These parameters are used to develop personalized car-following and lane-changing models. A joint SUMO-Python simulation platform validates the proposed models, demonstrating their ability to realistically replicate car-following and lane-changing behaviors across varying driving styles. Simulation results confirm the model’s accuracy and reliability, with traffic volumes closely reflecting actual traffic flow characteristics.
He et al. (Fri,) studied this question.