Geometric feature extraction and its profile accuracy evaluation for pockets on curved aircraft skin from scanned point clouds

Abstract

Dimension inspection is crucial for aircraft skin manufacturing and assembly. To meet weight loss and other functional requirements, large numbers of shallow pockets are distributed on the surface of thin-walled aircraft skin. It is difficult to evaluate the profile accuracy of the pockets, as the aircraft skin is composed of curved surfaces and its actual geometry is inconsistent with the nominal model. This study proposes a method to evaluate the profile and position accuracy for the pockets on aircraft skin utilizing point clouds obtained by 3D scanners. Firstly, the feature contour lines are identified and extracted from the scanned aircraft skin surface point clouds by adopting the local curvature-aware method and minimum bounding rectangle algorithm. Then, an algorithm is proposed to mitigate the contour identification deviations of the segmented area by employing the shortest geodesic distance and transition zone section projection, so as to enhance the measurement accuracy. Finally, the viability of the proposed method for evaluating pockets on aircraft skin is verified by the simulation, and the reliability and robustness of the proposed method are verified by the simulation and experiment. Experimental results show that the measurement accuracy of the pocket can be improved from 0.5mm to 0.0505mm, compared with traditional specialized conformation fixture methods.