A Geometric Digital Twin Modeling Method for Thin-Walled Microstructures Considering Laterally Distributed Error

Abstract

Precision microstructures are the core devices of electromechanical products, and their accuracy directly affects the product’s overall performance. Laterally thin-walled structures such as precision tooth plates and silicon arms are the most common types of precision microstructures. Due to the characteristics of their tiny size and complex shape, the laterally thin-walled surface is challenging to measure and model, which in turn makes it impossible to realize the accurate prediction and quantitative control of its assembly accuracy. To address this problem, this paper proposes a geometric digital twin modeling method considering the laterally distributed error of thin-walled microstructures. A geometric digital twin model that accurately represents the geometric distributed error of the structure’s lateral surfaces is built by applying an image measuring instrument to measure the point clouds of the upper and lower surface profiles, followed by point cloud filtering and merging based on a common datum, and finally surface reconstruction and integration with an ideal 3D model. The method is effectively validated on the silicon arm, a vital component of the thermal mechanical package. Its geometric modeling error is less than 1/10 of its tolerance, and the relative error between the model-based stiffness simulation calculation results and the experimental data is less than 10%. The method can effectively analyze the influence of the manufacturing errors of thin-walled lateral surfaces on the assembly accuracy and provides an adequate model basis for the control of the geometrical and physical performances of precision microstructures.