A Pareto-Based Estimation of Distribution Algorithm for Solving Multiobjective Distributed No-Wait Flow-Shop Scheduling Problem with Sequence-Dependent Setup Time
Weishi Shao,Dechang Pi,Zhongshi Shao
DOI: https://doi.org/10.1109/tase.2018.2886303
IF: 6.636
2019-01-01
IEEE Transactions on Automation Science and Engineering
Abstract:Influenced by the economic globalization, the distributed manufacturing has been a common production mode. This paper considers a multiobjective distributed no-wait flow-shop scheduling problem with sequence-dependent setup time (MDNWFSP-SDST). This scheduling problem exists in many real productions such as baker production, parallel computer system, and surgery scheduling. The performance criteria are the makespan and the total weight tardiness. In the MDNWFSP-SDST, several identical factories are considered with the related flow-shop scheduling problem with no-wait constraints. For solving the MDNWFSP-SDST, a Pareto-based estimation of distribution algorithm (PEDA) is presented. Three probabilistic models including the probability of jobs in empty factory, two jobs in the same factory, and the adjacent jobs are constructed. The PWQ heuristic is extended to the distributed environment to generate initial individuals. A sampling method with the referenced template is presented to generate offspring individuals. Several multiobjective neighborhood search methods are developed to optimize the quality of solutions. The comparison results show that the PEDA obviously outperforms other considered multiobjective optimization algorithms for addressing MDNWFSP-SDST. Note to Practitioners—This paper is motivated by the process cycles in multiproduction factories (or lines) of baker production, surgery scheduling, and parallel computer systems. In these process cycles, jobs are assigned to multiproduction factories (or lines), and no interruption exists between consecutive operations. This paper models this process as a multiobjective distributed no-wait flow-shop scheduling with SDST. Scheduling becomes more challenging when facing distributed factories. This paper provides an estimation of distributed algorithm with Pareto dominate concept which uses a probabilistic model to generate offspring. Experiment results suggest that the proposed algorithm can find superior solutions of large-scale instances. This scheduling model can be extended to practical problems by considering other constraints, such as assembly process, mixed no-wait, and transporting times. Besides, the proposed algorithm can be applied to solve other distributed scheduling problems and industrial cases, once their constraints are known, i.e., the processing time of operations, the setup time of machines.