Impact Factor:

8.0

DOI number:

10.1016/j.engappai.2026.115190

Affiliation of Author(s):

College of Shipbuilding Engineering, Harbin Engineering University

Journal:

Engineering Applications of Artificial Intelligence

Place of Publication:

England

Funded by:

National Natural Science Foundation of China (No. 52371315); Key Research and Development Program of

Key Words:

Cargo and research ports; Research vessels; Marine experiments; Uncertainty; Robust optimization; Differential evolution

Abstract:

Cargo and Research Ports (CRPs) involve a highly coupled scheduling problem that coordinates cargo vessels, research vessels, and marine experiments, while uncertainties in quay crane efficiency and marine experiment duration further increase operational complexity. Existing deterministic approaches are not suitable for handling such coupled uncertainty in CRP scheduling. To address this issue, this study proposes a robust optimization model, termed ROCREA, based on box uncertainty sets to jointly optimize vessel turnaround time, quay crane movement distance, and marine experiment completion time under uncertainty. A Multi-distribution Adaptive Hybrid Differential Evolution (MAHDE) algorithm is further developed to solve this model. By integrating multi-distribution parameter generation, success-history-based adaptation, and dynamic dual-strategy selection, MAHDE improves search diversity, adaptability, and solution stability for large-scale robust scheduling problems. Computational experiments based on synthetic test instances reflecting the practical operating conditions of CRPs in southern China show that ROCREA consistently maintains a 0% failure rate and outperforms the deterministic model under different uncertainty and congestion conditions. Meanwhile, compared with the benchmark algorithms considered in this study, MAHDE is competitive on small-scale instances and becomes significantly superior on medium-scale and large-scale scheduling problems, while still generating high-quality schedules for large-scale instances (70 vessels and 14 experiments) within 12 min. These results indicate that the proposed framework provides effective decision support for CRP scheduling under uncertainty.

Note:

National Natural Science Foundation of China (No. 52371315); Key Research and Development Program of Hainan Province (ZDYF2023GXJS017); Hainan Provincial Natural Science Foundation (525MS110).

Co-author:

Ming-Wei Li,Wei-Chiang Hong

First Author:

Xiang-Yang Li

Indexed by:

Journal paper

Correspondence Author:

Zhong-Yi Yang

Document Code:

115190

Discipline:

Engineering

Document Type:

J

Volume:

179

ISSN No.:

0952-1976

Translation or Not:

no

Date of Publication:

2026

Included Journals:

SCI

Links to published journals:

https://www.sciencedirect.com/science/article/pii/S0952197626014740