Intelligent design and buckling experiment of curvilinearly stiffened thin-walled structures.

• For the complex variable-stiffness features of curvilinear paths, non-uniform layouts and variable thicknesses, a unified image characterization method is proposed for the curvilinearly stiffened structures. • A combination strategy of intelligent and gradient optimization has been proposed as an approach to efficiently obtain different types of excellent load-bearing structural forms. • For the curvilinearly stiffened panel, the typical buckling experiments and corresponding FEA have been carried out, exploring the excellent load-bearing mechanism. Curvilinearly stiffened variable-stiffness structures offer excellent load-bearing capacities and design flexibility, particularly for thin-walled structures with cutouts. Nevertheless, the complex combination of numerous variables, buckling failure modes, and unclear load-carrying mechanisms present significant challenges to structural analysis and optimization. Hence, a unified characterization method for nonuniform layouts and variable thicknesses is proposed and an intelligent optimization framework based on structural image learning is established. Buckling experiments on two stiffened panels and a corresponding numerical analysis are performed to verify the effectiveness of the method under axial compressive loading. Results show that the curvilinearly stiffened panel exhibits a superior load-bearing capacity of 28.8% compared with an orthogonally stiffened panel. Specifically, the curvilinearly stiffened panel efficiently transfers axial loads to a broader region and avoids interruptions in the force transmission path by cutouts, thereby resulting in material and buckling failures simultaneously. [ABSTRACT FROM AUTHOR]