A CT Defect Detection Method for Additive Manufacturing Workpieces Driven by Digital Models and Virtual Simulation

Additive manufacturing (AM) technology has been widely applied in key fields such as aerospace due to its high design freedom and material utilization rate. However, internal defects of parts are difficult to be captured by traditional non-destructive testing methods. Industrial computed tomography (CT) can provide three-dimensional visualization of parts, offering a high-precision means for defect detection. Nevertheless, the scarcity and high acquisition cost of real defect sample data limit the training and application of deep learning. To address this issue, an industrial CT defect detection method for AM workpieces driven by CAD digital models and virtual simulation is proposed. Firstly, typical internal pore defects are parameterizedly introduced into a defect-free CAD model to form a defective digital model. Then, CT projection and reconstruction simulations are carried out to generate a large-scale simulated CT dataset. The generated simulated data can be used not only for comparing defect-free simulations with real CT data to assist manual and traditional algorithm defect identification but also as a training set for deep learning models, improving the accuracy and generalization ability of defect detection. A case study of a 3D resin-printed blade part was conducted to verify the effectiveness of the proposed method. The simulation and real scanning results were compared and analyzed, and the application prospects of the method were discussed, providing a new technical approach for non-destructive testing of AM.