<![CDATA[The Vehicle Routing Problem with Time Windows (VRPTW) is a challenging NP-hard problem in logistics optimization. This study evaluates a Swarm-Genetics algorithm, a hybrid method combining Particle Swarm Optimization (PSO) and Genetic Algorithm (GA) with swarm regeneration and adaptive parameter control. The algorithm was tested on 57 Solomon benchmark instances (C, R, RC) under three random seeds to assess robustness. Results show that the algorithm is robust across seeds, producing stable outcomes with minimal variation. It frequently preserves fleet efficiency, often matching the Best Known Solutions (BKS) in vehicle count, particularly for clustered instances. However, routing distances remain less competitive, with average gaps of about 10% for clustered, 12–13% for random, and over 20% for mixed cases. Convergence analysis further indicates rapid early improvements but stagnation in complex distributions. Overall, Swarm-Genetics provides a robust and fleet-efficient framework, though further enhancements are needed to improve distance quality.]]>
Copyrights © 2025
