Solving multi-objective hybrid flowshop lot-streaming scheduling with consistent and limited sub-lots via a knowledge-based memetic algorithm
All workpieces of a job are usually treated as a whole in general hybrid flowshop scheduling problem, resulting in lower production efficiency and on-time delivery rate. If a job lot can be split into smaller sub-lots and machines in each stage can process them in parallel, these performance indicat...
Saved in:
| Published in: | Journal of manufacturing systems Vol. 73; pp. 106 - 125 |
|---|---|
| Main Authors: | , , , , , |
| Format: | Journal Article |
| Language: | English |
| Published: |
Elsevier Ltd
01.04.2024
|
| Subjects: | |
| ISSN: | 0278-6125 |
| Online Access: | Get full text |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | All workpieces of a job are usually treated as a whole in general hybrid flowshop scheduling problem, resulting in lower production efficiency and on-time delivery rate. If a job lot can be split into smaller sub-lots and machines in each stage can process them in parallel, these performance indicators can be easily improved. Hence, this work addresses the hybrid flowshop lot-streaming scheduling problem with consistent sub-lots where the number of sub-lots is strictly limited by that of tools, and presents a new mixed-integer linear programming model to minimize the makespan and due time deviation simultaneously. To solve the problem efficiently, a knowledge-based memetic algorithm is proposed with the following four improvements. Firstly, four initialization knowledge produces high-quality initial solutions by making each decision variable located in the high-quality interval. Secondly, a multi-segment preservative crossover enhances the exploration ability by disturbing the population to reproduce new individuals. Thirdly, the self-adaptive selection mechanism saves computational resources and improves evolutionary efficiency by applying the historical success rates of six mutations to operator selection. Finally, deviation neutralization and critical path shortening neighborhood search heuristics based on problem-specific knowledge improve the exploitation ability by directional mining of promising solution spaces. The results of 4800 experiments show that the knowledge-based initialization methods produce the best initial solutions in 97% of cases, while the simultaneous use of the above four improvements can effectively improve the performance of the algorithm. The proposed algorithm also outperforms the six state-of-the-art algorithms under three stopping criteria in 16,800 experiments. The key managerial insights gleaned from the research findings are also underscored, together with the algorithm's limitations.
•A MILP model is developed for multi-objective HFLSP with consistent and limited sub-lots.•Four initialization knowledge are abstracted to produce high-quality initial solutions.•A self-adaptive selection of six mutation operators improves evolutionary efficiency.•DNR/CPSR-based neighborhood search heuristics enhance the exploitation ability.•Statistical results indicate that KMA outperforms all other comparison algorithms. |
|---|---|
| ISSN: | 0278-6125 |
| DOI: | 10.1016/j.jmsy.2024.01.006 |