Automatic precision matching method of optical measurement data for blade profile quality detection

Abstract

Blades are key components in aviation engines, determining energy conversion efficiency, operational safety, and lifespan. During the precision machining, assembly, and maintenance stages of damaged blades, three-dimensional measurements are necessary to accurately guide related process flows. To precisely measure various surface quality errors on blades, a measurement reference standard is usually introduced when aligning measurement data with CAD models. However, the complexity of thin-walled blades and significant noise interference makes it challenging to accurately locate optical measurement data. This paper proposes an automatic method for matching blade optical measurement data with CAD models, allowing for precise alignment of actual blade measurement data with their theoretical CAD models. To achieve this, the method prioritizes matching the tenon surface data based on structural features in the point cloud data. Firstly, a point feature descriptor is developed for the quick and automatic retrieval of tenon points from the blade point cloud. Secondly, a new sample consensus pose estimation technique is introduced. This method imposes constraints on the maximum distance from the normal vectors during the compatibility sorting process, facilitating the generation of more reasonable hypotheses. Through rigorous hypothesis verification, the optimal rotation and translation matrices are determined. Finally, experimental results on actual measurement data and simulated datasets demonstrate that the proposed matching method achieves high accuracy and efficiency.