Multipopulation for Multiobjective-Based Ant Colony System for Berth Allocation Problem

doi:10.1109/tits.2026.3713499

Multipopulation for Multiobjective-Based Ant Colony System for Berth Allocation Problem

2026-09-24

0

PRE

AI

DOI：10.1109/tits.2026.3713499

Original

Original request for help

Save

Abstract

En 中文

The berth allocation problem (BAP) is a crucial problem in the maritime transportation system that greatly affects the operating efficiency of container terminals. As an NP-hard combinatorial optimization problem, the efficiency and total service costs of BAP relate to multiple factors, such as the total waiting time, total handling time, and total delayed departure time of vessels. Therefore, the BAP in real-world situations is a complex multiobjective optimization problem, which poses great challenges to existing heuristic algorithms. To efficiently solve the multiobjective BAP, this article proposes a multipopulation for multiobjective-based ant colony system (MPMOACS) algorithm. The major contributions include three aspects. First, MPMOACS employs three colonies to optimize three objectives based on the efficient multiple populations for multiple objectives (MPMO) framework, so as to search all three objective spaces sufficiently. Second, a two-stage solution construction method is designed to firstly perform vessel-to-berth allocation for vessels and then vessel-to-vessel sequence arrangement at berths, so as to reduce the search complexity. Third, an elite learning strategy combining small- and large-scope learning is designed to further approximate the Pareto front and improve global optimization ability. Experimental results show that, in almost all of the tested instances with various scales, MPMOACS outperforms the ACS algorithm, other cutting-edge evolutionary computation (EC) algorithms for BAPs, and some widely used and state-of-the-art multiobjective algorithms, in terms of each objective, the integrated objective, and the hypervolume metric of the obtained solution sets. To be specific, the improvement is up to 34.26% compared to the other EC algorithms for BAP in the instance of the largest scale.

Keywords:

Berth allocation problem

multiobjective optimization problem

evolutionary computation

ant colony system

multiple populations for multiple objectives

Multipopulation for Multiobjective-Based Ant Colony System for Berth Allocation Problem

Abstract

Journal

Organization

Cited Papers