In this study, the distributed flexible assembly permutation flowshop scheduling problem (DFAPFSP) with sequence-dependent setup times and makespan criterion was investigated. The DFAPFSP comprises two distinct phases: a distributed permutation flowshop in the initial stage, followed by an integration phase. The integration stage, which employs multiple parallel assembly machines, achieves significantly higher throughput efficiency compared to monolithic assembly machine architectures in high-volume manufacturing scenarios. A novel mixed-integer linear programming model was established to describe the problem. The DFAPFSP can be divided into two stages: production and assembly. A two-stage iterated greedy (TSIG) algorithm was designed based on the two-stage characteristics of the DFAPFSP. In the first stage, the production plan is optimized, and in the second stage, the assembly plan is optimized. The destruction, reconstruction, and local search algorithms in the two stages and acceptance criterion were redesigned. Numerous computational experiments and performance evaluations were performed by comparing the TSIG algorithm with state-of-the-art algorithms. The results and discussions show that the proposed TSIG algorithm is better than its peers for solving the DFAPFSP.

In this paper, distributed permutation flowshop problems with a weighted total completion time criterion (DPFSP-WTC) were addressed to minimize the completion time of all factories. First, the completion time of all factories were converted to a single one by a novel strategy, and a mixed integer programming model was developed. Second, an improved iterated greedy (IIG) algorithm was proposed. Based on features of the concerned problems, a simple heuristic is designed to improve the quality of initialization solutions. A local search operation was developed to improve the convergence performance of the proposed algorithm. Finally, numerous experiments were carried out for solving 720 instances with different scales. The proposed IIG was compared with five state-of-the-art algorithms. The comparisons and discussions showed that the proposed IIG has superior performance compared to its peers.