Weld-seam identification and model reconstruction of remanufacturing blade based on three-dimensional vision. (August 2021)

• Record Type: Journal Article
• Authors: Jing, Ou; Lai, Zou; Qinghong, Wan; Xin, Liu; Yingjie, Li
• Is Part Of: Advanced engineering informatics. Volume 49(2021)
• Publication Date: 2021-08
• Subjects: Binocular vision -- Three-dimensional reconstruction -- Point cloud denoising -- Normal filtering -- Remanufacturing blade -- Weld-seam; Computer-aided engineering -- Periodicals; Engineering -- Data processing -- Periodicals; 620.00285
• DOI: 10.1016/j.aei.2021.101300
• Languages: English
• ISSNs: 1474-0346

Abstract

Highlights: The partial reconstruction of the blade is achieved based on stereo vision and image processing method to obtain 3D data of weld seam efficiently. The point cloud normal filtering method and region growth algorithm are combined to precisely segment weld-seam point cloud. Abstract: Remanufacturing technique has been widely used for repairing the damaged regions of aero-engine blades. As the irregularity of weld-seam shape and uneven residual height pose great challenges for the subsequent precision grinding and polishing, it is necessary to use the high efficiency and precision measurement method to obtain a true and accurate weld-seam model, especially in the boundary areas. Therefore, a measurement approach for weld-seam identification and model reconstruction of the remanufacturing blade based on self-developed binocular vision system is presented in this work. The calculation of three-dimensional reconstruction and the complexity of subsequent point cloud processing were reduced by coarse positioning firstly, and then the weld-seam was located precisely by point cloud segmentation based on the region growth algorithm and point cloud normal filtering method. On this basis, a true weld-seam model with precise boundary was obtained. The average error of the boundary extracted by proposed method is about 0.263 mm lower than that of the traditional method. In addition, the B-spline surface was fitted according to the point cloud without weld-seam feature, and a theoretical machining model was obtained by cutting of B-spline surface along the weld-seam boundary. Furthermore, the results of verification experiment indicated that the smoothness error of machined blade surface was less than 0.025 mm and the machining error was less than 0.07 mm. This method has obvious advantages in calculation efficiency and reconstruction accuracy of theoretical machining model while compared with traditional measurement methods for the remanufacturing blade, which is beneficial to improve the machining quality and efficiency of the remanufacturing parts.