A novel allowance evaluation method of blade based on high-precision matching and deviation calculating for 3D points

Abstract

3D points are increasingly used in evaluating allowance of complex curved blade. However, the error of position-attitude alignment and dimensional deviation calculation between nominal and measured models will reduce accuracy of blade allowance. Therefore, a novel allowance evaluation method is proposed to enhance allowance evaluation accuracy by overcoming above problems. In this method, a point cloud registration algorithm is developed by the inlier correspondences selection methods (PCR-ICS) based on local features of 3D points and rigid geometric consistency. For different sets of point cloud, the registration accuracy with this algorithm is improved by over 30% than that with some typical registration algorithms. Considering unsmooth and missing regions of the measured model, a novel point-to-plane deviation calculation (PTPDC) algorithm is developed by constructing planes in conical region of normal vector of each point in the nominal model. The effectiveness of proposed method is verified by GOM scanning experiments of blades with different sizes and shapes. The allowance evaluation error of these blades all is less than 0.03 mm compared to the results obtained by coordinate measuring machine.