The digital twin-enabled deformation prediction and control methodologies for thin-walled parts in intelligent manufacturing: a review

Thin-walled parts are widely used in the aerospace industry for their lightweight properties and high aerodynamic efficiency. However, they are prone to significant deformation during machining, which greatly affects product quality. Effective prediction and control strategies for machining deformation are crucial for improving machining quality and stability. This paper systematically reviews research on deformation prediction and control for thin-walled parts enabled by digital twins. While traditional physics-based methods provide interpretability, they lack adaptability, whereas purely data-driven approaches excel in nonlinear fitting but are hampered by data dependency and poor generalisation. Consequently, a significant gap exists between these paradigms. This review makes a substantial contribution by critically analysing this dichotomy and advocating for a hybrid modelling framework as a pathway forward. We systematically synthesise how digital twins facilitate the integration of physical mechanisms with data-driven models, such as PINNs, to achieve accurate, real-time, and reliable deformation management. Furthermore, the paper outlines the challenges and future prospects of this hybrid intelligent control approach in intelligent manufacturing.