Influence of Structural and Material Uncertain Parameters on Honeycomb Sandwich Panels Based on Neural Networks

Due to the thin-walled nature of the aluminum foil, the performance of honeycomb sandwich panels is susceptible to the influence of processing and the inherent discreteness of the material itself, resulting in fluctuations. Existing studies have very limited research on the uncertainty of the out-of-plane compressive performance of honeycomb sandwich panels. This research aims to propose a method for modeling the uncertainty of honeycomb sandwich panels, which can be used to analyze and quantify the impact of material and structural uncertainties on the out-of-plane mechanical properties of honeycomb sandwich panels, thereby filling the aforementioned research gap. A finite element model capable of expressing the uncertainties of structure and material was established. A backpropagation (BP) neural network model was established to predict the out-of-plane mechanical properties of honeycomb sandwich panels. Thereafter, the extended Fourier amplitude sensitivity test method (EFAST) and Monte Carlo simulation were applied to identify and quantify the key uncertain parameters. The results indicate that aluminum foil elastic modulus and thickness-to-span ratio are the main parameters that affect the out-of-plane modulus. Angle of double cell wall, aluminum foil yield strength, and thickness-to-span ratio are the main parameters that affect the out-of-plane strength. Simultaneous ± 5% variations in all uncertain parameters lead to variations in the out-of-plane modulus and strength of [-9.96%, 10.61%] and [-10.34%, 17.06%], respectively. The coefficient of variations is 4.3% and 5.1%, respectively. Increasing foil thickness and using higher-yield-strength materials can improve honeycomb panel stability, but will also shift performance boundaries. Therefore, under uncertainty, design must holistically account for the dispersion, mean, and bound of performance variation.