Digital Twin in Smart Manufacturing
He Zhang, Zitong Wang, and Fei Tao1,2**
Digital Twin International Research Center, International Institute for Interdisciplinary and Frontiers, Beihang University, Beijing, China 2 School of Automation Science and Electrical Engineering, Beihang University, Beijing, China
E-mail: ftao@buaa.edu.cn
Status
The idea of the digital twin could be traced to the Apollo 13 mission in the 1960s in which multiple simulators were employed to evaluate the failure, train astronauts and mission controllers in response to the oxygen tank explosion. After a period of dormancy, digital twin re-emerged in the 21st century and attracted widespread attention. Prof. Grieves proposed a three-dimensional model of the digital twin and expounded on its value and significance in the full life cycle management of products [1]. NASA listed it as one of the key paths in its future development blueprint [2]. Prof. Tao proposed a five-dimension digital twin model which contains physical objects, virtual models, data, connections, and services to further promote the practice of digital twins [3]. To date, the digital twin has been applied into multiple fields, and the smart manufacturing is one of the most popular fields because it aligns with the core of Industry 4.0, that is, to achieve seamless integration of vertical and horizontal information flows in the supply chain and value chain through digital technology, and to build a highly intelligent production system [4,5]. And the digital twin has been applied into various aspects in smart manufacturing and revolutionized the traditional manufacturing mode. Although some companies or researchers have carried out the practice of digital twin in smart manufacturing, the current maturity of digital twin application is still not high enough to fully utilize the advantages and value of digital twin due to the limitations of cognitive understanding as well as technology. In addition, in recent years, Industry 5.0, which emphasizes on human-centeredness, sustainability and resilience, has been proposed, putting new requirements on the development of digital twins [6]. And some advanced technologies, such as Large Language Model (LLM), have advanced by leaps and bounds in recent years, bringing new opportunities for digital twin development [7]. In this context, digital twins still need to be further developed to improve its intelligence level, maturity and application scale.
Current and future challenges
Current and future challenges of digital twin in smart manufacturing contains many aspects, such as application scenarios, key technologies, and security. Current industrial application scenarios of digital twins predominantly focus on real-time condition monitoring, quality prediction, and intelligent control in automobiles, airplanes, ships, and other fields. However, research on digital twins in extreme manufacturing,
e.g., microfabrication, ultra-precision manufacturing, and giant-systems manufacturing, is still at a relatively blank stage. And the current level of digital twins is also difficult to handle for extremely complex systems or projects. Data is a key driver for digital twins in manufacturing [8]. With the development and advancement of sensor and communication technologies, more and more manufacturing process data can be captured [9]. However, on one hand, transient anomalies that may arise during the manufacturing process are still difficult to capture. On the other hand, manufacturing data remains difficult to collect in extreme environments. In addition, the massive data collected from sensors and controllers, combined with that generated by digital twin models, poses a significant challenge for rapid analysis and processing due to limited computing capability Models are one of the important foundations for realizing digital twins in smart manufacturing. However, the current digital twin models are still constructed as one-off solutions tailored to specific use cases, which
limits their generalizability [10]. The requirement for related domain knowledge further limits their application and development. While some scholars have explored the use of purely data-driven digital twins, this approach based on black-box algorithms poses interpretability challenges. Once a problem arises, it is difficult to effectively allocate responsibility. The implementation of digital twins in smart manufacturing is not possible without the support of related software or platforms. Currently, some companies have developed related tools such as Ansys Twin Builder, Azure Digital Twins and 3DEXPERIENCE. However, there is insufficient compatibility between the different software. The functionality of each piece of software is also insufficient to support the entire digital twin chain in multiple scenarios across different fields [11].
Advances in science and technology to meet challenges
To address these challenges, significant scientific and technological innovations are emerging across multiple domains. These advances aim to enhance model accuracy, data interoperability, computational efficiency, and security, which could further enable scalable industrial applications. The combination of systems engineering thinking and complexity theory with digital twins is a promising approach for applying digital twins to more complex objects and scenarios. Systems engineering frameworks such as Model-Based Systems Engineering (MBSE) contribute to unifying multi-domain models and digital threads, thereby enabling collaborative intelligent manufacturing. Moreover, the complexity science would be helpful for addressing nonlinear dynamics inherent in large-scale industrial systems through multi-scale analysis and complex networks, offering tools to enhance the resilience [12]. To address latency and computational bottlenecks, hybrid edge-cloud architectures are being deployed. The increasing processing power of smart chips at the edge helps to achieve low latency and high real-time data transmission, enhanced data privacy and security, reduced bandwidth consumption, and lower costs [13]. Furthermore, as quantum computing technology develops and matures, it can help form clusters to work together to process larger data sets [14]. Advances in artificial intelligence (AI) are transforming the way of digital twin modelling. Currently, advanced algorithms, such as Physics Informed Neural Networks (PINNs), are integrating domain knowledge more deeply to improve interpretability [15,16]. Besides, generative AI would further enable synthetic data generation to fill gaps in training datasets, enhancing predictive maintenance accuracy. In the future, enabled by generative AI, automatic generation of complex digital twin models based on user requirements is also possible [17]. And blockchain-based traceability solutions are being implemented to mitigate security risks [18]. Standardized frameworks such as ISO 23247 named Automation systems and integration — Digital twin framework for manufacturing are addressing data silos [19]. Tools such as Amazon IoT TwinMaker and Eclipse Ditto enable cross-platform integration through modular APIs and universal asset models. In addition, makeTwin, a unified reference architecture for digital twin software platform, has been proposed [11]. However, the related international standards should be further developed to improve compatibility. Collaboration among all relevant stakeholders is also important for the formation of a digital twin industrial software ecosystem. Technological advancements, such as AI, edge-cloud collaboration, blockchain, and standardization, are collectively addressing the core challenges of digital twins in smart manufacturing. Continued innovation in quantum computing, explainable AI, and cross-industry collaboration will further accelerate the adoption of digital twin applications.
Concluding remarks
Digital twins have become a cornerstone of Industry 4.0 and 5.0, facilitating the integration of physical and virtual systems in smart manufacturing. Despite their superiority in real-time monitoring, predictive
analytics, and intelligent control, challenges persist in applications to complex scenarios, such as extreme manufacturing environments, ensuring data integrity, and overcoming computational and interoperability limitations. Collaborative innovation across industries, coupled with robust policy frameworks, will unlock the full potential of digital twins, transforming them from reactive tools into proactive enablers of next-generation industrial intelligence.
Acknowledgements
The work was supported by Beijing Natural Science Foundation under Grant L243009, National Natural Science Foundation of China under Grant 52120105008, and China Postdoctoral Science Foundation under Grant 2024M754054.
References
[1] Grieves M 2014 Digital Twin: Manufacturing Excellence Through Virtual Factory Replication (White Paper) 1-7 [2] Glaessgen E and Stargel D 2012 The digital twin paradigm for future NASA and US Air Force vehicles 53rd AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials Conference 20th AIAA/ASME/AHS Adaptive Structures Conference 14th AIAA 1818 (doi:10.2514/6.2012-1818) [3] Tao F, Zhang M, Liu Y et al 2018 Digital twin driven prognostics and health management for complex equipment CIRP Ann 67 169-172(doi:10.1016/j.cirp.2018.04.055) [4] Tao F and Qi Q 2019 Make more digital twins Nature 573490-491(doi:10.1038/d41586-019-02849-1) [5] Tao F, Zhang H and Zhang C 2024 Advancements and challenges of digital twins in industry *Nat. Comput. Sci.*4169- 177(doi:10.1038/s43588-024-00603-w) [6] Zhang H, Li Y, Zhang S et al 2025 Artificial Intelligence-Enhanced Digital Twin Systems Engineering Towards the Industrial Metaverse in the Era of Industry 5.0 *Chin. J. Mech. Eng.*3840(doi:10.1186/s10033-025-01210-0) [7] Gautam A, Aryal M R, Deshpande S et al 2025 IIoT-enabled digital twin for legacy and smart factory machines with LLM integration *J. Manuf. Syst.*80511-523(doi:10.1016/j.jmsy.2025.03.022) [8] Friederich J, Francis D P, Lazarova-Molnar S et al 2022 A framework for data-driven digital twins of smart manufacturing systems *Comput. Ind.*136103586 [9] Zhang M, Tao F, Huang B et al 2024 Digital twin data: methods and key technologies Digit. Twin 12 (doi:10.12688/digitaltwin.17467.2) [10] Kapteyn M G, Pretorius J V R and Willcox K E 2021 A probabilistic graphical model foundation for enabling predictive digital twins at scale *Nat. Comput. Sci.*1337-347(doi:10.1038/s43588-021-00069-0) [11] Tao F, Sun X, Cheng J et al 2024 makeTwin: A reference architecture for digital twin software platform *Chin. J. Aeronaut.*371- 18 (doi:10.1016/j.cja.2023.05.002) [12] Grieves M 2024 Intelligent digital twins and the development and management of complex systems Digit. Twin 18 (doi:10.12688/digitaltwin.17574.1) [13] Alcaraz C and Lopez J 2022 Digital twin: A comprehensive survey of security threats *IEEE Commun. Surv. Tutor.*241475- 1503(doi:10.1109/COMST.2022.3171465) [14] Gyongyosi L and Imre S 2019 A survey on quantum computing technology *Comput. Sci. Rev.*3151-71 (doi:10.1016/j.cosrev.2018.11.002) [15] Kobayashi K and Alam S B 2024 Explainable, interpretable, and trustworthy AI for an intelligent digital twin: A case study on remaining useful life *Eng. Appl. Artif. Intell.*129107620(doi:10.1016/j.engappai.2023.107620) [16] Kim S, Choi J H and Kim N H 2022 Data-driven prognostics with low-fidelity physical information for digital twin: physics-informed neural network *Struct. Multidiscip. Optim.*65255(doi:10.1007/s00158-022-03348-0) [17] Mata O, Ponce P, Perez C et al 2025 Digital twin designs with generative AI: crafting a comprehensive framework for manufacturing systems *J. Intell. Manuf.*11-24(doi:10.1007/s10845-025-02583-8) [18] Putz B, Dietz M, Empl P et al 2021 Ethertwin: Blockchain-based secure digital twin information management Inf. Process. *Manag.*58102425(doi:10.1016/j.ipm.2020.102425) [19] Yoo S K, Sun K J and Kim S H 2024 Digital Twin Standardization: Trends and Future Prospects Electron. Telecommun. Trends 3979-86(doi: 10.22648/ETRI.2024.J.390308)