Maritime route optimization using Q-learning–based quantum-behaved particle swarm optimization

The rapid growth of global maritime transportation has intensified demands for safe and efficient shipping routes optimization. Conventional particle swarm optimization (Particle Swarm Optimization PSO) and its quantum variant (Quantum-behaved Particle Swarm Optimization QPSO) frequently encounter challenges such as premature convergence and generate infeasible solutions under complex navigational constraints. To overcome these limitations, we propose a novel Q-learning–based quantum-behaved Particle Swarm Optimization (QLQPSO) algorithm. This approach incorporates a multi-dimensional state space and a Q-learning controller to adaptively regulate the contraction–expansion balance, attraction-center switching, simulated annealing, and path smoothing. Furthermore, a customized “projection + smoothing” operator is introduced to ensure navigational feasibility of the generated routes. Experimental evaluations conducted on the China-Australia shipping corridor demonstrate that the QLQPSO algorithm achieves superior performance, including the shortest voyage distance (3772 nm), reduced curvature, and near-perfect feasibility, consistently outperforming PSO, QPSO, Reinforcement Learning Particle Swarm Optimization(RLPSO), and Reinforcement Learning Quantum-behaved Particle Swarm Optimization(RLQPSO). Ablation studies further verify the critical roles of the Q-learning-driven scheduling mechanism and the constraint-repair operat or in enhancing algorithm effectiveness. In summary, the QLQPSO algorithm presents a robust and efficient solution for maritime route optimization in complex nautical environments, offering significant potential for practical applications in the maritime industry.