Pusan National University Unveils Bayesian Calibration Breakthrough for Digital Twins in Semiconductor Manufacturing
Pusan National University Unveils Bayesian Calibration Breakthrough for Digital Twins in Semiconductor Manufacturing
A new Bayesian calibration framework from Pusan National University aims to boost prediction accuracy for digital twins used in semiconductor material handling systems.
BUSAN, South Korea, July 22, 2025 – As semiconductor and display manufacturing becomes more complex, automated material handling systems (AMHSs) are essential. While digital twins help manage and optimize these systems, discrepancies between digital models and real-world systems can hinder performance and create delays.
The newly developed Bayesian calibration framework tackles both parameter uncertainty and discrepancy—two critical factors that impact the reliability of digital twins for automated material handling in semiconductor and display industries. This approach significantly enhances decision-making and improves production outcomes.
Digital twins of AMHSs are challenged by parameter uncertainty—stemming from hard-to-measure real-world data—and discrepancy, which results from operational differences between the physical system and its digital counterpart. Over time, these issues reduce predictive accuracy, yet most calibration techniques address only parameter uncertainty, often requiring large amounts of field data and ignoring discrepancy.
A research team led by Professor Soondo Hong from the Department of Industrial Engineering at Pusan National University, South Korea, introduced a Bayesian calibration method to solve these challenges. “Our framework optimizes calibration parameters while compensating for discrepancies,” explained Prof. Hong. “It can scale to large smart factory setups and delivers robust calibration with far less field data than typical methods.” The findings were published online on May 8, 2025, and appeared in Volume 80 of the Journal of Manufacturing Systems on June 1, 2025.
The team implemented modular Bayesian calibration across multiple operating conditions, enabling estimation of uncertain parameters and adjustment for discrepancies using minimal real-world input. This combines observational and prior knowledge with digital twin simulations through probabilistic models—namely, Gaussian processes—yielding a posterior distribution for calibrated predictions. Three models were evaluated:
- A surrogate using only field data to predict system behavior;
- A standard digital twin using calibrated parameters;
- And a calibrated digital twin accounting for both uncertainty and discrepancy.
The calibrated twin model outperformed the field-only surrogate and provided better prediction accuracy than the standard digital twin. Prof. Hong emphasized, “Our approach makes effective calibration possible even with limited real-world observations, addressing model discrepancies directly. It’s a flexible, validated method adaptable to each facility.”
This calibration system offers a practical solution to optimize digital twins where scale, complexity, or flexibility are required. It accurately predicted large-scale system responses with limited observations and sped up calibration for future production schedules. The method is particularly valuable for digital models that diverge from real-world behavior due to simplified logic or code. Industries with complex material handling—where manual optimization is impractical—can leverage this framework for sustainable, reusable digital twin models that can be transferred across sectors. The framework is currently in deployment at Samsung Display, where it is being customized for operational realities with the support of on-site teams.
Looking ahead, Prof. Hong said, “Our research paves the way for self-adaptive digital twins and, in the future, could be a key enabler of smart manufacturing.”
More info here – Have a Story? Address it to the Editor and submit it here
About Pusan National University
Founded in 1946, Pusan National University (PNU) is one of South Korea’s leading national research universities, located in Busan. Renowned for its commitment to innovation and academic excellence, PNU offers a wide range of undergraduate, graduate, and doctoral programs across disciplines such as engineering, natural sciences, humanities, medicine, and business. The university is recognized for its world-class research in smart manufacturing, artificial intelligence, advanced materials, and digital twin technology. PNU maintains robust partnerships with industry leaders, government agencies, and international institutions to drive cutting-edge research and global collaboration.
With a diverse student body and faculty, PNU is dedicated to fostering future leaders equipped with the skills and knowledge needed to address complex challenges in today’s rapidly evolving technological landscape. Pusan National University’s main campus is located at 2 Busandaehak-ro 63beon-gil, Geumjeong-gu, Busan, South Korea.
Disclaimer
The information provided in this article is for general informational purposes only and from publicly available sources. While we strive for accuracy, we do not make any representations or warranties, express or implied, regarding the completeness, reliability, or validity of the content. This article does not make any direct claims about specific companies, individuals, or organizations. Any references to reports or external sources are for context and do not imply endorsement or verification of any specific allegations. Readers are encouraged to conduct their own research and seek professional advice before making business decisions. We disclaim any liability for any losses or damages incurred as a result of reliance on the information provided.