A bi-objective evolutionary algorithm for distributed production scheduling with eligibility restrictions

• Published in: Applied soft computing 2025-03, Vol.171, p.112738, Article 112738
• Main Authors: Hu, Chaoming, Zhang, Teng, Gao, Yao, Liu, Xinbao, Wang, Xubiao
• Format: Article
• Language: English
• Subjects: Customer satisfaction; Distributed production scheduling; Eligibility restrictions; Fuzzy transportation durations; GA-VNS algorithm
• ISSN: 1568-4946
• DOI: 10.1016/j.asoc.2025.112738

The distributed manufacturing model has emerged as a highly effective strategy in the manufacturing sector, esteemed for its prowess in cutting enterprise operating costs and elevating market responsiveness. This paper focuses on a new energy vehicle battery manufacturer and provides a comprehensive exploration of the distributed production scheduling process. Particular emphasis is placed on factory eligibility restrictions and the inherent fuzziness of transportation durations. Our objectives are twofold: to minimize total costs, encompassing both production and transportation expenses, while simultaneously maximizing overall customer satisfaction. Given the stringent eligibility criteria, orders are strictly confined to designated factories. Customer satisfaction, a vital metric, is significantly influenced by the alignment between the fuzzy arrival time and the due time of an order. This alignment is often shaped by consumer sensitivity and transportation duration's inherent uncertainties. To tackle this intricate challenge, we propose a bi-objective approach that integrates the genetic algorithm and variable neighborhood search (GA-VNS). Experimental results reveal that in over two-thirds of the test instances, the ratio of non-dominated solutions (Rnd) of the GA-VNS algorithm exceeds 80 %, and in all instances, its average distance (AD) to the reference set is less than 0.03, demonstrating significant advantages over NSGA-II, MOABC, and MOWOA algorithms. •Distributed production scheduling with eligibility restrictions and fuzzy transportation durations.•Two key objectives considered are the total cost and the level of customer satisfaction.•A bi-objective approach that combines genetic algorithm and variable neighborhood search (GA-VNS) is developed.•The proposed GA-VNS algorithm outperforms alternative bi-objective algorithms, including NSGA II, MOABC, and MOWOA.