Additive manufacturing error quantification on stability of composite sandwich plates with lattice-cores through machine learning technique

Architected lattice structures are designed to achieve specific mechanical properties while maintaining relative lightweight. Additive Manufacturing (AM) facilitates the fabrication of lattice structures with complex geometries. However, manufacturing imperfections, e.g., node dislocation, radius variation, and waviness, inevitably affect the performance of composite structures, of which the impact is significant yet difficult to quantify. Herein, a virtual model-assisted AM error quantification scheme is proposed to alleviate this challenge. The influence of geometric imperfections on the static buckling behaviour for sandwich lattice-core panels is investigated. A recently developed Extended Support Vector Regression (X-SVR) is utilized to alternatively bridge multiscale analyses. By integrating the sampling and virtual modelling methods, the effect of AM errors can be comprehensively quantified with statistical moments, probability density function (PDF), cumulative distribution function (CDF), etc. Furthermore, high computational efficiency, robustness, and other inherent features highlight the applicability of the proposed AM error quantification scheme in engineering.