The vibrations and noises caused by traffic, machine operation, and construction activity have become a serios concern. Fortunately, the use of elastic metamaterials proposes new methods for reducing those vibration. But there are still issues with rapid and accurate design periodic barrier system based on vibration characteristics. To this end, a dual deep learning network model (DDLNM) is developed to inversely design the periodic barrier system consisted of three components (TCPB) for reducing those vibrations in this paper. Seven variables, containing material and geometric features of unite cell of the TCPB for damping vibrations, are considered. Besides, the designed parameters of the TCPB for reducing the vibrations caused by trains are obtained according to the measured vibration characteristics by adopting the DDLNM. To verify correction and effectiveness of the designed TCPB for reducing vibrations, a two-dimensional (2D) finite element model (FEM) is established. The results exhibit that the designed TCPB according to the DDLNM can effectively mitigate the vibrations caused by trains, and the attenuation ratio can be up to 51.4%. This further demonstrates that the proposed dual network model based on deep learning has good rapidity, accuracy, and universality on inversely designing the periodic barriers based on vibration features.
Inverse Design and Application Periodic Barriers for Isolating Ambient Vibration Based on Deep Learning
KSCE J Civ Eng
KSCE Journal of Civil Engineering ; 28 , 9 ; 3601-3615
2024-09-01
15 pages
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch