Low-frequency vibration bandgaps and deep learning-based intelligent design method of Y-shaped core sandwich metabeams

In this paper, the wave propagation mechanics theoretical model of Y-shaped core sandwich metabeams with local resonators is established based on the spectral element method, and the flexural wave bandgaps and vibration isolation characteristics are studied. The reliability of the theoretical model of Y-shaped core sandwich metabeams is verified by the finite element method and experiment. On this basis, a deep learning-based intelligent design method for predicting the vibration transmission characteristics and structure design of Y-shaped core sandwich metabeams is proposed. A dataset is created using the theoretical model of wave dynamics, and deep neural networks are constructed to forward prediction of transmission characteristics and intelligent design of Y-shaped core sandwich metabeams, respectively. Results indicate that the bandgap range of Y-shaped core sandwich metabeams is 2.5 times that of Y-shaped core sandwich beams, and the start frequency of bandgap is reduced by 578 Hz. The prediction of the transmission characteristic curve is in good agreement with the target, and the average relative error of intelligent design is below 3 %, which verifies the precision of the intelligent design method. The innovative, intelligent design method provides a novel way to realize engineering vibration reduction design of acoustic metamaterials rapidly