Repair volume extraction method for damaged parts in remanufacturing repair

Abstract

The remanufacturing repair of damaged parts provides great potential for restoring them to a like-new condition. Existing methods cannot build an effective connection between 3D point cloud reconstruction and laser metal deposition (LMD) repair, which results in a low degree of integration and automation through the entire remanufacturing repair process. In this paper, we propose a repair volume extraction method that integrates surface reconstruction and repair volume extraction. For the surface reconstruction, a few reference points reconstructed by a close-range photogrammetry system (XJTUDP) were used to perform the initial registration and coordinate system transformation. Then, the internal and external parameters of two cameras embedded into a 3D optical scanning system (XJTUOM) were estimated using an eight-step calibration method. After completing the reconstruction of local point clouds through XJTUOM, we utilized another initial registration strategy depending on reference points and fine registration via the ICP algorithm to determine a refined complete point cloud for the damaged part. For the repair volume extraction, the refined complete point cloud was converted from the XJTUDP coordinate system to the LMD coordinate system, and two types of initial registration approaches and the ICP algorithm were used to achieve a best-fit position between the refined complete point cloud and the nominal point cloud. Finally, a point cloud representation of the repair volume was extracted by a distance-based filtering operation. The proposed method is validated by two experiments designed to extract the repair volume of a damaged gear with both planar and non-planar broken surfaces and is proven to be effective, robust, and highly automated.