Hybrid Digital Twin Framework with Meta-Learning and Reinforcement Learning for Nonlinear Manufacturing Systems
DOI:
https://doi.org/10.21467/proceedings.7.4.4Keywords:
Hybrid Digital Twin, Meta-Learning, Nonlinear Manufacturing SystemsAbstract
The rapid pace of change in non-linear manufacturing systems, especially in high-stakes industries like semiconductors and pharmaceuticals, demands fresh approaches to boost efficiency and adaptability. This research establishes a hybrid framework of meta-learning and reinforcement learning to address the concept of hybrid digital twin demands. The innovative approach aims to enhance real-time adaptability to tackle nonlinear complexities in manufacturing. The digital twins provide active and real-time models of physical processes that could be monitored and optimized easily; in semiconductor manufacturing, these devices combine precision and speed. This approach strengthens traditional models to empower adaptive learning aspects, modifying themselves to fit changing conditions quickly in situations in which rapid production changes are called for due to market developments and technology enhancements. But again, in pharmaceutical manufacturing, where compliance with regulations and product quality are crucial, this process promotes preemptive decision making supported by predictive modeling and real-time data analysis. This hybrid algorithm reduces downtime through meta-learning to speed up adapting to new data and through reinforcement learning to continuously optimize the process. It increases yield rates and enables a more robust production process. From current tests, it has demonstrated to be faster and more responsive than traditional practices in addressing complex manufacturing needs. Through the simulation of various "what-if" scenarios, they also provide manufacturers with a means to test and hone without the exposure to high costs. Programs aimed at smart manufacturing could deliver further innovation and sustainable developments if advanced algorithms were applied. Recognizing a hybrid digital twin in manufacturing will have a huge impact on the acceleration of industries in coming years.
References
[1] M. Zhang and J. Chen, “Hybrid machine learning approaches for semiconductor process control,” Journal of Process Control, vol. 76, pp. 163–176, Apr. 2019, doi: 10.1016/j.jprocont.2019.04.002.
[2] C. Finn, P. Abbeel, and S. Levine, “Model-agnostic meta-learning for fast adaptation of deep networks,” in Proceedings of the 34th International Conference on Machine Learning (ICML), Sydney, Australia, Aug. 2017, pp. 1126–1135.
[3] H. Chen, T. Li, and X. Zhang, “Meta-learning for intelligent manufacturing systems: A comprehensive survey,” Journal of Manufacturing Systems, vol. 59, pp. 481–509, Jul. 2021, doi: 10.1016/j.jmsy.2021.03.015.
[4] D. Silver et al., “A general reinforcement learning algorithm that masters chess, shogi, and Go through self-play,” Science, vol. 362, no. 6419, pp. 1140–1144, Dec. 2018, doi: 10.1126/science.aar6404.
[5] R. Kumar and M. Shen, “Thermal process optimization in semiconductor manufacturing using deep reinforcement learning,” IEEE Transactions on Semiconductor Manufacturing, vol. 32, no. 4, pp. 416–425, Nov. 2019, doi: 10.1109/TSM.2019.2933731.
[6] B. Bagheri, T. Yang, and C. Lu, “Digital twin framework for reconfigurable manufacturing systems: Design and implementation,” Journal of Manufacturing Systems, vol. 54, pp. 157–171, Jan. 2020, doi: 10.1016/j.jmsy.2020.01.003.
[7] F. Tao, Q. Qi, and A. Y. C. Nee, “Digital twin-driven product design framework,” International Journal of Production Research, vol. 57, no. 12, pp. 3935–3953, Jun. 2019, doi: 10.1080/00207543.2019.1579932.
[8] S. Choi and D. Lee, “Real-time optimization for semiconductor manufacturing: A review,” Computers and Chemical Engineering, vol. 112, pp. 174–189, Mar. 2018, doi: 10.1016/j.compchemeng.2018.01.014.
[9] J. Lee, S. Kim, and H. Park, “Digital twin for semiconductor manufacturing: A review,” IEEE Access, vol. 8, pp. 137847–137862, Jul. 2020, doi: 10.1109/ACCESS.2020.3012287.
[10] Y. Wang, L. Gao, and M. Wang, “Reinforcement learning for semiconductor fabrication: A review,” IEEE Transactions on Semiconductor Manufacturing, vol. 33, no. 2, pp. 283–296, May 2020, doi: 10.1109/TSM.2020.2971450.
[11] M. Wang, J. Zhang, and K. Zhou, “Hybrid control strategies for semiconductor manufacturing processes: A comprehensive review,” IEEE Transactions on Industrial Electronics, vol. 67, no. 4, pp. 2928–2936, Apr. 2020, doi: 10.1109/TIE.2019.2916393.
Downloads
Published
Issue
Section
License
Copyright (c) 2025 Harshinni B (Author)
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
