TY - JOUR
T1 - Solving the dynamic energy aware job shop scheduling problem with the heterogeneous parallel genetic algorithm
AU - Luo, Jia
AU - El Baz, Didier
AU - Xue, Rui
AU - Hu, Jinglu
N1 - Funding Information:
This work is supported in part by the Japan Society for the Promotion of Science. Moreover, Didier El Baz is grateful to NVIDIA Corporation for the donation of the Tesla K40 GPUs used in this work and the authors would like to express their gratitude to the editors and the reviewers for their helpful comments.
Funding Information:
This work is supported in part by the Japan Society for the Promotion of Science . Moreover, Didier El Baz is grateful to NVIDIA Corporation for the donation of the Tesla K40 GPUs used in this work and the authors would like to express their gratitude to the editors and the reviewers for their helpful comments.
Publisher Copyright:
© 2020 Elsevier B.V.
PY - 2020/7
Y1 - 2020/7
N2 - Integrating energy savings into production efficiency is considered as one essential factor in modern industrial practice. A lot of research dealing with energy efficiency problems in the manufacturing process focuses solely on building a mathematical model within a static scenario. However, in the physical world shop scheduling problems are dynamic where unexpected events may lead to changes in the original schedule after the start time. This paper makes an investigation into minimizing the total tardiness, the total energy cost and the disruption to the original schedule in the job shop with new urgent arrival jobs. Because of the NP hardness of this problem, a dual heterogeneous island parallel genetic algorithm with the event driven strategy is developed. To reach a quick response in the dynamic scenario, the method we propose is made with a two-level parallelization where the lower level is appropriate for concurrent execution within GPUs or a multi-core CPU while codes from the two sides can be executed simultaneously at the upper level. In the end, numerical tests are implemented and display that the proposed approach can solve the problem efficiently. Meanwhile, the average results have been improved with a significant execution time decrease.
AB - Integrating energy savings into production efficiency is considered as one essential factor in modern industrial practice. A lot of research dealing with energy efficiency problems in the manufacturing process focuses solely on building a mathematical model within a static scenario. However, in the physical world shop scheduling problems are dynamic where unexpected events may lead to changes in the original schedule after the start time. This paper makes an investigation into minimizing the total tardiness, the total energy cost and the disruption to the original schedule in the job shop with new urgent arrival jobs. Because of the NP hardness of this problem, a dual heterogeneous island parallel genetic algorithm with the event driven strategy is developed. To reach a quick response in the dynamic scenario, the method we propose is made with a two-level parallelization where the lower level is appropriate for concurrent execution within GPUs or a multi-core CPU while codes from the two sides can be executed simultaneously at the upper level. In the end, numerical tests are implemented and display that the proposed approach can solve the problem efficiently. Meanwhile, the average results have been improved with a significant execution time decrease.
KW - Dynamic scheduling
KW - Energy efficiency
KW - GPU computing
KW - Job shop scheduling
KW - Multi-core processing
KW - Parallel genetic algorithm
UR - http://www.scopus.com/inward/record.url?scp=85079879878&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85079879878&partnerID=8YFLogxK
U2 - 10.1016/j.future.2020.02.019
DO - 10.1016/j.future.2020.02.019
M3 - Article
AN - SCOPUS:85079879878
SN - 0167-739X
VL - 108
SP - 119
EP - 134
JO - Future Generation Computer Systems
JF - Future Generation Computer Systems
ER -