Multifeature-Fitting and Shape-Adaption Algorithm for Component Repair

Abstract

In recent years, the usage of additive manufacturing (AM) provides new capabilities for component repair, which includes low heat input, small heat-affected zone, and freeform near-net-shape fabrication. Because the geometry of each worn component is unique, the automated repair process is a challenging and important task. The focus of this paper is to investigate and develop a general best-fit and shape-adaption algorithm for automating alignment and defect reconstruction for component repair. The basic principle of using features for rigid-body best-fitting is analyzed and a multifeature-fitting method is proposed to best fit the 3D mesh model of a worn component and its nominal component. The multifeature-fitting algorithm in this paper couples the least-squares method and a density-based outlier detection method. These two methods run alternately to approach the best-fit result gradually and eliminate the disturbance caused from the defect geometry. The shape-adaption algorithm is used to do cross section comparison and defect reconstruction based on the best-fitted 3D model. A “point-line-surface” fracture surface detection method is proposed to construct fracture surface and the fracture surface boundary is dilated to trim the nominal 3D model to obtain defect geometry. Illustrative examples with typical components and different kinds of defects are used to demonstrate the flexibility and capability of using multifeature-fitting and shape-adaption algorithm developed in this paper.