Digital twin-based positioning error compensation method for double-slotted flap

• Digital twin framework integrates machine vision and GPR for flap control; • Vision-based PnP algorithm enables real-time joint pose estimation; • Composite kernel GPR captures hybrid error trends and micro-fluctuations; • Closed-loop strategy reduces aft flap positioning error by 91.6%. The kinematic accuracy of double-slotted flap mechanisms is critical to the aerodynamic performance of aircraft. However, physical prototypes inevitably suffer from positioning errors induced by manufacturing tolerances, assembly clearances, and gravitational deformation, causing the actual trajectory to deviate from the theoretical design. To address this challenge and achieve high-precision control, this paper proposes a closed-loop framework integrating digital twins (DT), machine vision, and Gaussian Process Regression (GPR). First, a machine vision system is established based on the Perspective-n-Point (PnP) algorithm and geometric constraints to enable real-time pose estimation and virtual-physical mapping of key joints. Second, to address the hybrid nature of errors—characterized by macroscopic trends mixed with microscopic fluctuations—a GPR model with a combined kernel function is designed for precise nonlinear error prediction and uncertainty quantification. Finally, an inverse mapping algorithm is implemented to convert predicted angular errors into motor compensation commands, forming an integrated "perception-prediction-correction" closed loop. Experimental results demonstrate that the system effectively identifies and compensates for positioning deviations. Compared with the uncompensated mode, the proposed strategy reduces the mean positioning error of the aft tab by approximately 91.6% and the maximum error by 88.8%, significantly enhancing the mechanism's positioning accuracy.