NDE Data Fusion between Inconsistent Geometries

Abstract

Data fusion in the NDE digital thread and digital twin requires spatial registration, but in many cases, as-designed and as-built geometries are different. Because of the additional prior knowledge of geometry and topology, a 3D CAD modeling context shows great potential to seamlessly integrate different kinds of NDE data in the digital thread. Non-Uniform Rational B-splines (NURBS) are the standard representation of surfaces in a CAD system. Recently, a differentiable NURBS framework was developed with mathematical formulations of the NURBS derivatives with respect to the input parameters. Using this framework, we perform gradient-based optimization. However, it is still challenging to transform the NDE data to a usable reference CAD geometry with a consistent as-built CAD model and a measured point cloud to support data fusion applications. We have tested this approach on a bar geometry as proof of concept. Our preliminary results show that the approach can fit the distorted surfaces to the measured NDE data well without introducing gaps between the surfaces of the CAD model.