A Q-learning based hyper-heuristic scheduling algorithm with multi-rule selection for sub-assembly in shipbuilding

Abstract

Sub-assembly is the basic stage of ship hull construction. It is necessary to optimize the scheduling of sub-assembly to shorten its assembly cycle and ensure the normal execution of subsequent processes. The scheduling problem of sub-assembly is an NP-hard problem that should take into consideration both spatial layout and temporal schedule. In this work, a mathematical model for scheduling the sub-assembly is established, and a Q-learning based hyper-heuristic with multi-spatial layout rule selection is proposed. Specifically, a spatial layout method based on multi-rule selection is proposed first. In various scenarios, distinct spatial layout rules are chosen to derive an appropriate spatial arrangement. Subsequently, a hyper-heuristic algorithm based on Q-learning is crafted to optimize the scheduling sequence and the selection of spatial layout rules. As a verification, numerical experiments are carried out in cases of different scales collected from a large shipyard. The effectiveness of the proposed algorithm is verified by comparing it with different spatial layout algorithms, various heuristic operators, existing well-known hyper-heuristic methods, and other Q-learning based scheduling methods. The results suggest that the proposed algorithm outperforms other comparison algorithms in most testing cases.