Bilevel Optimization Framework for Dynamic Accessibility-Driven Taxi Service Point Planning

Optimizing taxi service accessibility requires balancing strategic infrastructure decisions with operational flexibility. While facility location and dynamic dispatching have been studied separately, their hierarchical interdependence remains unexplored. This study develops a bilevel optimization framework where the upper level determines fixed service point locations, and the lower level dynamically allocates service areas based on temporal demand variations. We propose an improved Harris Hawks optimization (IHHO) algorithm with adaptive weighting mechanisms balancing service efficiency, load distribution, and spatial continuity. Experimental validation using Manhattan taxi data demonstrates statistically significant improvements of 5.4%–6.9% in mean performance (p<0.001; Cohen’s d>1.67) compared to traditional HHO, particle swarm optimization, and differential evolution algorithms. More significantly, IHHO achieves 20-fold improvement in load balancing and fourfold improvement in distribution stability, addressing operational requirements critical for practical deployment. The framework provides transportation planners with an integrated decision support tool, advancing hierarchical optimization methods in urban transportation systems.