Demand-Responsive Bus Route Planning Considering Passenger Behavior and Road Time Impedance

Abstract

This study tackles the "last-mile" connectivity problem at high-speed rail hubs by optimizing Demand-Responsive Transit (DRT) systems. A novel framework integrates the Adaptive Large Neighborhood Search (ALNS) algorithm, Prospect Theory, and time-dependent road network models. It balances operational costs and passenger satisfaction using soft time windows and loss-aversion-based penalties. Empirical analysis using taxi GPS data from Xi’an North Railway Station shows that dynamic insertion reduces unserved demand by 14% with only a 7.8% increase in operational cost. The proposed behaviorally informed, data-driven methodology significantly enhances nighttime service efficiency and provides a resilient feeder solution for major transport hubs.