Multi-Fidelity Robust Design Optimisation For Composite Structures Based On Low-Fidelity Models Using Successive High-Fidelity Corrections

In this paper, a novel mull-fidelity modelling-based optimisation framework is developed for the robust design of composite structures. The proposed framework provides significant savings on computation time compared to both conventional mull-fidelity and high-fidelity modelling methods while maintaining an acceptable level of accuracy. Artificial neural networks (ANNs) and mull-level optimisation approach are both incorporated into this mull-fidelity modelling formulation. The framework utilises varied High-Fidelity Model (HFM) and Low-Fidelity Model (LFM) covering different design spaces. This means that the HFM has only a few design variables, whereas the LFM explores the entire design spaces during the optimisation process. The proposed multi-fidelity formulation is demonstrated by the robust design optimisation (RDO) of a mono-stringer stiffened composite panel considering design uncertainty under non-linear post-buckling regime.