A macro-micro coupled evolution model of highway traffic flow considering driving behavior heterogeneity

Abstract

Highway traffic dynamics emerge from interactions between aggregate traffic states and individual driving behaviors, yet most models represent only one scale. This study proposes a bidirectional macro-micro coupled model integrating the Cell Transmission Model, the Intelligent Driver Model, and a MOBIL-inspired lane-changing mechanism. In the micro-tomacro pathway, low-speed vehicles amplify local congestion and lane changes redistribute density across lanes. In the macro-to-micro pathway, congestion reduces desired speeds, while inter-lane speed variance increases lane-changing motivation. Aggressive, normal, and conservative driver types are modeled using distinct behavioral parameters. Simulations on a 10 km, three-lane highway under free-flow, congestion, incident, aggressive-dominated, and conservative-dominated scenarios show that the coupled model reproduces traffic dynamics beyond those captured by purely macroscopic or microscopic models. Ablation results indicate that macro-to-micro feedback contributes about 65% of the total coupling effect, increasing lane-change rates by up to 12.5% relative to the micro-only model and producing clearer differences in congestion adaptation and incident-induced queue formation.