A Data-Driven Quasi-Dynamic Traffic Assignment Model Integrating Multi-Source Traffic Sensor Data on the Expressway Network

Static traffic assignment (STA) models have been widely utilized in the field of strategic transport planning. However, STA cannot fully represent dynamic road conditions and suffer from inaccurate during congestion. At same time, an increasing number installed sensors become important means detecting conditions. To address shortcomings models, we integrate multi-source sensor datasets propose a novel data-driven quasi-dynamic model, named DQ-DTA. In this records toll stations are used for time-varying travel demand estimation. GPS trajectory vehicles further to calculate link costs network, replacing imprecise Bureau Public Roads (BPR) function. Moreover, license plate recognition (LPR) data design statistical probability-based multipath method capture travelers’ route choices. The expressway network Hunan province is selected as study area, several classic also chosen performance comparison. Experimental results demonstrate that accuracy proposed DQ-DTA model about 6% higher than models.