Open Access

ARTICLE

Multi-Level Subpopulation-Based Particle Swarm Optimization Algorithm for Hybrid Flow Shop Scheduling Problem with Limited Buffers

Yuan Zou¹, Chao Lu^1,*, Lvjiang Yin², Xiaoyu Wen³

1 School of Computer Sciences, China University of Geosciences, Wuhan, 430074, China
2 School of Economics and Management, Hubei University of Automotive, Shiyan, 442002, China
3 Henan Key Laboratory of Intelligent Manufacturing of Mechanical Equipment, Zhengzhou University of Light Industry, Zhengzhou, 450002, China

* Corresponding Author: Chao Lu. Email:

(This article belongs to the Special Issue: Applications of Artificial Intelligence in Smart Manufacturing)

Computers, Materials & Continua 2025, 84(2), 2305-2330. https://doi.org/10.32604/cmc.2025.065972

Received 26 March 2025; Accepted 13 May 2025; Issue published 03 July 2025

Abstract

The shop scheduling problem with limited buffers has broad applications in real-world production scenarios, so this research direction is of great practical significance. However, there is currently little research on the hybrid flow shop scheduling problem with limited buffers (LBHFSP). This paper deeply investigates the LBHFSP to optimize the goal of the total completion time. To better solve the LBHFSP, a multi-level subpopulation-based particle swarm optimization algorithm (MLPSO) is proposed, which is founded on the attributes of the LBHFSP and the shortcomings of the basic PSO (particle swarm optimization) algorithm. In MLPSO, firstly, considering the impact of the limited buffers on the process of subsequent operations, a specific circular decoding strategy is developed to accommodate the characteristics of limited buffers. Secondly, an initialization strategy based on blocking time is designed to enhance the quality and diversity of the initial population. Afterward, a multi-level subpopulation collaborative search is developed to prevent being trapped in a local optimum and improve the global exploration capability. Additionally, a local search strategy based on the first blocked job is designed to enhance the MLPSO algorithm’s exploitation capability. Lastly, numerous experiments are carried out to test the performance of the proposed MLPSO by comparing it with classical intelligent optimization and popular algorithms in recent years. The results confirm that the proposed MLPSO has an outstanding performance when compared to other algorithms when solving LBHFSP.

---

Keywords

---