Macro-meso-scale fusion modeling based stochastic multi-scale heterogeneous nested method for buckling reliability analysis of the composite structure

The multi-scale differences in materials lead to the uncertain failure behavior of the structure. This paper proposes a reliability analysis method based on the stochastic multi-scale heterogeneous nesting (SMHN) method, transforming building block testing into building block modeling and analysis. First, a key contribution is the Parallel Dynamic Dimensionality Reduction (PDDR) method, which uses experimental CFRP data to resolve multi-scale mapping conflicts by dynamically retaining discriminative features across scales, reducing information transmission burden while efficiently predicting anisotropic elastic parameters; Secondly, a multi-scale reliability analysis framework with SMHN as the core is established, and the data-driven mesoscopic model is embedded in the physical simulation-driven macroscopic model to realize heterogeneous nesting; In the case verification, the buckling reliability of the stiffened panel was taken as the research object, combined with PDDR to improve the modulus prediction accuracy (Relative Standard Deviation Error, RSDE) by 62.646%, and the direct reliability evaluation considering the high-dimensional mesoscopic characteristic parameters in complex structures was realized based on SMHN, which provided a new solution to the multi-scale reliability problem.