Abstract: The hierarchical stiffened shell is one kind of innovative aerospace thin-walled structures, and it has the advantages of lightweight and high load-carrying capacity. Since the stiffener configurations of hierarchical stiffened shells are complex, they would result in longer computational time of post-buckling analysis. In this case, it is meaningful to improve the efficiency of post-buckling analysis and optimization for the rapid design of hierarchical stiffened shells. Variable-Fidelity Model (VFM) has been widely used in the design and optimization of complex engineering structures. High-Fidelity Model (HFM) and Low-Fidelity Model (LFM) linked by the bridge function can indicate high prediction accuracy and low computational cost. In this paper, the establishment methods of HFM and LFM are proposed for hierarchical stiffened shells. Then, the Gaussian process regression method is employed to establish VFM, where an adaptive updating method is proposed according to the root mean square error of VFM. Results indicate that, when achieving the similar prediction accuracy, the computational cost of the VFM based on the proposed method is lower by about 60% than that of the surrogate model based on the direct sampling of HFM, indicating the significant advantage of the proposed method in prediction accuracy. In addition, the effects of various types of LFMs on the prediction accuracy of VFM are discussed. Results reveal that, with regard to the post-buckling problems for the load-carrying capacity prediction of hierarchical stiffened shells, the prediction accuracy of VFM can be improved if the post-buckling analysis ability is retained in the establishment of LFM.

Key words: variable-fidelity model, hierarchical stiffened shells, buckling analysis, adaptive updating method, asymptotic homogenization method