PLANNING FOR SELECTIVE REMARSHALING IN AN AUTOMATED CONTAINER TERMINAL USING COEVOLUTIONARY ALGORITHMS

Remarshaling in a container terminal refers to the task of rearranging containers stored in the stacking yard to improve the efficiency of subsequent loading onto a vessel. When the time allowed for such preparatory work is limited, only a selected subset of containers can be rearranged. This paper proposes a cooperative co-evolutionary algorithm (CCEA) that decomposes the planning problem into three subproblems of selecting containers, determining target locations, and finding a moving order, and conducts a cooperative parallel search to find a good solution for each subproblem. To cope with the uncertainty of crane operation in real terminals, the proposed method iteratively replans at regular intervals to minimize the gap between the plan and the execution. For an efficient search under real-time constraint of iterative replanning, our CCEA reuses the final populations of the previous iteration instead of restarting from scratch. Significance: This paper deals with an optimization problem having three constituent subproblems that are not independent of each other. Instead of solving the subproblems in turn and/or heuristically, which sacrifices solution quality for efficiency, we employ a CCEA to conduct a cooperative parallel search to find a good solution efficiently. For applications to real world, issues like real-time constraint and uncertainty are also addressed.