Digital twin technology facilitates precision improvement in complex product assembly: A progressive deduction method of data-driven tolerance allocation

Advancements in aerospace technology have significantly increased the demands on assembly precision, particularly for large aircraft models. These developments require sub-millimeter precision to be achieved in assembly, posing new challenges for conventional tolerance allocation methods, which rely heavily on theoretical models and static simulations. To address these limitations, this paper proposes a digital twin system for online deduction of assembly tolerances (DTS-ODTA) using point cloud registration and free-form deformation (FFD). The proposed method includes three key innovations. First, a digital twin (DT)-based assembly process control framework supports the simulation of assembly deviations by integrating virtual and actual data. Second, a data-driven method models the actual surface shapes, ensuring synchronization between physical and digital states. Third, a progressive deduction strategy for tolerance allocation combines virtual and actual information, providing effective inputs for real-time assembly process control. This approach was validated through an industrial case study, achieving a prediction accuracy of 91.64 % for assembly precision and effectively preventing out-of-tolerance issues during product assembly. By promptly identifying and addressing potential assembly discrepancies, this method enhances the overall assembly quality and provides critical support for design decisions. In summary, this paper introduces a novel DT-based tolerance allocation method that shifts from traditional open-loop design to closed-loop control, thereby meeting the high precision demands of modern aerospace manufacturing.