Feature driven optimization design for free-form bulge-formed joint based on Kriging model

<title>Abstract</title> <p>This paper proposes a feature layering-based forward-reverse modeling method aimed at optimizing free-form bulge-formed joints. Beginning with the physical model of the joint, surface features are decomposed into six key global and local shape parameters on the basis of feature layering to establish an initial parametric surface, which is then discretized into point clouds and reconstructed into a Non-Uniform Rational B-Splines (NURBS) model utilizing the Least Square Progressive and Iterative Approximation (LSPIA) algorithm. An orthogonal experimental design is employed to sample the parameter space. Subsequently, a Kriging model is constructed to capture the nonlinear relationships between the shape parameters and joint performance, including mechanical properties and fairness. Through comparison with polynomial response surface models and Support Vector Machine (SVM) models the accuracy of the Kriging model is demonstrated, which is thereby facilitating the optimization of the joint surface design. The optimized free-form surface model demonstrates the effectiveness of the proposed method and provides a basis for designing similar free-form surface structures.</p>