An efficient multi-fidelity ant colony system for multi-crane scheduling optimization

초록

The yard crane scheduling problem (YCSP) of the shipbuilding industry aims to schedule blocks from the stack yard to the dock yard effectively, which is a key optimization problem in the shipbuilding process and is widely en countered in actual scenarios. This paper formulates a mathematical model for the yard crane scheduling problem with extended tandem scheduling (YCSP-T), a crucial and common process but often overlooked in the existing literature. To solve the YCSP-T, we propose an efficient Multi-Fidelity Ant Colony System (MF-ACS) algorithm. The algorithm fuses low-fidelity and high-fidelity modules using a priority partition strategy to enhance search efficiency and reduce the computationally expensive burden. In addition, a specialized synchronous tandem block control strategy is designed to handle the constraints in YCSP-T. The experimental results on real-world scenar ios demonstrate that the proposed MF-ACS algorithm outperforms several well-performing existing algorithms, including the work-balanced genetic algorithm and the multiobjective analytical evolutionary algorithm, achiev ing a 7%-13% reduction in travel distance. Moreover, it also exhibits a significant advantage in computational efficiency, achieving an acceleration rate of 2-5 times.

키워드