In the rapidly evolving landscape of modern manufacturing and engineering, a new technology is emerging as a crucial enabler-Data-Model Fusion (DMF). A recent review paper published in Engineering delves into this technology, exploring its methods, applications, and future prospects.
Model-based and data-driven methods are two important approaches in smart manufacturing and digital engineering. Model-based methods rely on physical laws or prior domain knowledge, while data-driven methods analyze data to extract insights. However, both methods have limitations. Model-based methods may suffer from low computational accuracy and high computational burdens, and data-driven methods often face issues such as poor interpretability and redundant features.
DMF offers a solution by integrating these two methods. It combines the advantages of model-based and data-driven methods, reducing the data-dependency of data-driven methods and enhancing the interpretability of data-driven models. DMF can be classified into four levels: data-level, feature-level, method-level, and decision-level. Each level represents a different way of integrating model-based and data-driven methods.
The paper also presents a conceptual framework for DMF, which includes data-driven methods, model-based methods, fusion strategies, services, and connections. This framework helps to clearly define the objectives and constraints involved in deploying DMF.
DMF has a wide range of applications in the product lifecycle, including product design, manufacturing, experimentation, testing, verification (ETV), and maintenance. In product design, it can optimize design processes and reduce computational costs. In manufacturing, it is used for process control, improving efficiency and productivity. In product maintenance, DMF enhances the interpretability of data-driven methods and improves the accuracy of predictions. In ETV, it reduces the computational complexity by using surrogate models and hybrid models.
Looking ahead, the future of DMF is promising. Multidisciplinary theories such as digital engineering and digital twins will guide its development, providing a more comprehensive information space and enabling more accurate decision-making. Emerging technologies like large language models and cloud-edge-end collaboration will support its implementation, reducing computational costs and improving the allocation of computational tasks. Moreover, DMF has potential applications in areas such as manufacturing service collaboration and virtual testing, which can further improve the efficiency of manufacturing systems.
DMF is a technology with great potential in smart manufacturing and digital engineering. As research and development in this area continue, we can expect to see more innovative applications and improvements in industrial processes.
The paper “Data–model Fusion Methods and Applications toward Smart Manufacturing and Digital Engineering,” authored by Fei Tao, Yilin Li, Yupeng Wei, Chenyuan Zhang, Ying Zuo. Full text of the open access paper: https://doi.org/10.1016/j.eng.2024.12.034.