Insider Brief
Researchers have developed a new Bayesian calibration framework that significantly improves the accuracy of digital twin models for automated material handling systems (AMHSs) by addressing both parameter uncertainty and system discrepancy.
The framework uses sparse field data and probabilistic modeling to calibrate digital twins, outperforming conventional models and enabling faster, more reliable predictions in complex manufacturing environments.
The method has been validated through empirical testing, applied at Samsung Display, and is designed to scale across various industries seeking accurate, self-adaptive digital twin solutions.
PRESS RELEASE — Digital twins for automated material handling systems (AMHSs) of semiconductor and display fabrication industries suffer from parameter uncertainty and discrepancy. This leads to inaccurate predictions, ultimately affecting performance. To address this, researchers have developed a new Bayesian calibration framework that simultaneously accounts for both parameter uncertainty and discrepancy, improving the prediction accuracy of digital twin models. This innovative framework holds great potential for enhancing digital twin applicability across diverse industries.
To manage increasingly complex manufacturing systems, involving material flows across numerous transporters, machines, and storage locations, the semiconductors and display fabrication industries have implemented automated material handling systems (AMHSs). AMHSs typically involve complex manufacturing steps and control logic, and digital twin models have emerged as a promising solution to enhance the visibility, predictability, and responsiveness of production and material handling operation systems. However, digital twins don’t always fully reflect reality, potentially affecting production performance and may result in delays.
Digital twins of AMHSs face two major issues: parameter uncertainty and discrepancy. Parameter uncertainty arises from real-world parameters that are difficult to measure precisely but are essential for accurate modeling. For example, the acceleration of an automated vehicle in AMHSs can vary slightly in the field but is fixed in the digital twin. Discrepancy, on the other hand, originates from the difference in operational logic between the real-world system and the digital twin. This is especially important since digital twins typically simplify or resemble the real processes, and discrepancies accumulated over time lead to inaccurate predictions. Despite its importance, most performance-level calibration frameworks overlook discrepancy and focus only on parameter uncertainty. Moreover, they often require a large amount of field data.
To address this gap, a research team led by Professor Soondo Hong from the Department of Industrial Engineering at Pusan National University, South Korea, developed a new Bayesian calibration framework. “Our framework enables us to simultaneously optimize calibration parameters and compensate for discrepancy,” explains Prof. Hong. “It is designed to scale across large smart factory environments, delivering reliable calibration performance with significantly less field data than conventional methods.” Their study was made available online on May 08, 2025, and published in Volume 80 of the Journal of Manufacturing Systems on June 01, 2025.
The researchers applied modular Bayesian calibration for various operating scenarios. Bayesian calibration can use sparse real-world data to estimate uncertain parameters while also accounting for discrepancy. It works by combining field observations and available prior knowledge with digital twin simulation results through probabilistic models, specifically Gaussian processes, to obtain a posterior distribution of calibrated digital twin outcomes over various operating scenarios. They compared the performance of three models: a field-only surrogate that predicts real-world behavior directly from observed data; a baseline digital twin model using only calibrated parameters; and the calibrated digital twin model accounting for both parameter uncertainty and discrepancy.
The calibrated digital twin model significantly outperformed the field-only surrogate and showed concrete improvements in prediction accuracy over the baseline digital models. “Our approach enables effective calibration even with scant real-world observations, while also accounting for inherent model discrepancy.” notes Prof. Hong, “Importantly, it offers a practical and reusable calibration procedure validated through empirical experiments, and can be customized for each facility’s characteristics.”
The developed framework is a practical and reusable approach that can be used to accurately calibrate and optimize digital twins, otherwise hindered by scale, discrepancy, complexity, or the need to be flexible for widespread cross-industry application. This approach accurately predicted field system responses for large-scale systems with scarce field observations and supported rapid calibration of future production schedules in real-world systems. The calibration system is also apt for discrepancy-prone digital models that behave differently than their real-world counterparts due to simplified logic or code. High-complexity production and material handling environments, where manual optimization is challenging, can also benefit from this calibration framework. It also enables the development of reusable and sustainable digital twin frameworks that can be applied to different industries. Furthermore, this approach is being applied and scaled at Samsung Display, where the researchers have closely collaborated with operation teams to customize the framework for the real-world complexities.
Overall, this novel framework has the potential to change the applicability and effectivity of AMHSs. Looking ahead, Prof. Hong concludes, “Our research offers a pathway toward self-adaptive digital twins, and in the future, has strong potential to become a core enabler of smart manufacturing.”
