Mathematical Modeling and Optimization of Energy-Conscious Flexible Job Shop Scheduling Problem With Worker Flexibility

• Published in: IEEE access Vol. 7; pp. 68043 - 68059
• Main Authors: Meng, Leilei, Zhang, Chaoyong, Zhang, Biao, Ren, Yaping
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 2019; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Decoding; dual-resource; Energy; Energy consumption; Energy-saving; Flexibility; flexible job shop scheduling; Genetic algorithms; Integer programming; Job shop scheduling; Job shops; Linear programming; Machine tools; Mixed integer; mixed integer linear programming; Optimization; Single machine scheduling; Strategy; variable neighbourhood search
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2019.2916468

This paper addresses the dual-resource constrained flexible job shop scheduling problem (DRCFJSP) with minimizing energy consumption. It is the first to study the energy-conscious DRCFJSP with turn OFF/ ON strategy. Different from the classical FJSP, the worker flexibility is considered in DRCFJSP. First, in order to solve this problem, we propose two mixed integer linear programming (MILP) models based on two modeling ideas, namely, idle time and idle energy. Because DRCFJSP is NP-hard, then we propose an efficient variable neighborhood search (VNS) algorithm. In the proposed VNS algorithm, eight neighborhood structures are designed to generate neighboring solutions. In addition, four energy-saving decoding approaches are specifically designed, in which two energy-saving strategies, namely, postponing strategy and turn OFF/ ON strategy are designed. Finally, the MILP model, the energy-conscious decoding methods, and the VNS are evaluated on numerical tests, whose effectiveness is shown by the experimental results. The experimental results show that the MILP model based on idle energy performs better than the model based on idle time idea, and the greedy hybrid decoding method outperforms the other three decoding methods. Moreover, the proposed VNS with eight neighborhood structures is a very competitive algorithm for the energy-conscious DRCFJSP.