An accuracy and performance-oriented accurate digital twin modeling method for precision microstructures.

Digital twin, a core technology for intelligent manufacturing, has gained extensive research interest. The current research was mainly focused on digital twin based on design models representing ideal geometric features and behaviors at macroscopic scales, which is challenging to accurately represent accuracy and performance. However, a numerical representation is essential for precision microstructures whose accuracy and performance are difficult to measure. The concept of a digital twin for an accurate representation, proposed in 2015, is still in the conceptual stage without a clear construction method. Therefore, the goal of accurate representation has not been achieved. This paper defines the concept and connotation of an accuracy and performance-oriented accurate digital twin model and establishes its architecture in two levels: geometric and physical. First, a geometric digital twin model is constructed by the contact surfaces distributed error modeling and virtual assembly with nonuniform contact states. Then, based on this, a physical digital twin model is constructed by considering the linear and nonlinear response of the structural internal physical properties to the external environment and time to characterize the accuracy and performance variation. Finally, the models are evaluated. The method is validated on microtarget assembly. The estimated values of surface modeling, center offset, and stress prediction accuracy are 94.22%, 89.3%, and 83.27%. This paper provides a modeling methodology for the digital twin research to accurately represent accuracy and performance, which is critical for product quality improvements in intelligent manufacturing. Research results can be extended to larger-scale precision structures for performance prediction and optimization. [ABSTRACT FROM AUTHOR]