The container stowage planning problem in a barge convoy system via Q-learning-based NSGA-II algorithm

The Container Stowage Planning Problem (CSPP) is a very complex and challenging problem involving the interests of shipping companies and ports. In inland shipping, multiple barges can be assembled into a fleet to transport a large number of containers to and from the hinterland. For this scenario, El Yaagoubi et al. [1] proposed the 3D-barge Convoy Container Stowage Planning Problem (3D-CCSPP), which is an extension of CSPP. This article optimizes the 3D-CCSPP mathematical model and proposes a Q-learning-based non-dominated genetic algorithm (Q-NSGA-II) to solve the model efficiently. The main innovation of the algorithm is its new mutation operator selection mechanism, which incorporates Q-learning to select the appropriate mutation operator during the search process. Comprehensive computational experiments were carried out on a wide range of examples to evaluate the performance of the proposed algorithm. Experimental results show that for the 3D-CCSPP problem, the proposed Q-NSGA-II can obtain a better set of non-dominated solutions. Moreover, the algorithm is more stable, with better convergence and robustness.