Optimizing a decomposed high-speed rail multi-station shortest sequence problem

• Published in: Expert systems with applications Vol. 283; p. 127543
• Main Author: Roy, Sandeepan
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 15.07.2025
• Subjects: Ant colony optimization; Decomposition; High-speed rail optimization; Multi-station location; Multi-swarm particle swarm optimization; Shortest sequence/route; Ant colony optimization; Decomposition; Multi-station location; High-speed rail optimization; Shortest sequence/route; Multi-swarm particle swarm optimization
• ISSN: 0957-4174
• DOI: 10.1016/j.eswa.2025.127543

High-speed rail (HSR) planning requires optimizing station locations and routes to balance utility, ridership, operational efficiency, and cost-effective construction. Existing studies often rely on single-heuristic and simultaneous optimization approaches that assume predetermined station sequences while overlooking real-world geographic constraints and utility factors. This study introduces a novel nonlinear optimization model to optimize both HSR multi-station locations and shortest sequence/routes (HSR-MSL-SS) and addresses key limitations of existing methods. This model integrates three main objectives: maximizing station location utility and ridership and minimizing overall corridor length, subject to station location and corridor constraints. The problem is decomposed or divided into two interrelated sub-problems: multi-station location selection and shortest sequence determination. A hybrid metaheuristic optimization algorithm integrating multi-swarm particle swarm optimization (MPSO) and ant colony optimization (ACO) is developed to solve these sub-problems iteratively. Two hybridization strategies are presented: Se-MPSO-ACO, which sequentially optimizes multi-station location and routing, and Si-MPSO-ACO, which optimizes both sub-problems simultaneously. As per the author’s knowledge, this is the first attempt at solving the HSR-MSL-SS problem using a decomposed and hybrid swarm intelligence-based approach. A real-world HSR case study validates the approach. The Se-MPSO-ACO and Si-MPSO-ACO achieved 10.8% and 7.8% improvement in objective function values, respectively, over the expert-designed alternative. Sensitivity analysis indicates that network connectivity and population accessibility have the most significant impact on total utility. Convergence analysis revealed that while both methods perform well, the Si-MPSO-ACO is more responsive to changes in weight factors and demonstrates greater stability towards convergence. Se-MPSO-ACO provides robust solutions with lower computational complexity.