| Abstrakt: |
Given the tight relationship between the effective material qualities of honeycomb cellular structures and their geometric cell configurations, these attributes play an important role in isolating specific target sound frequencies to prevent transmission through cellular panels. In this study, we introduce a predictive model to investigate the structural-acoustic performance of sandwich panels featuring honeycomb cellular cores, employing a combination of finite element (FE) analysis and machine learning-based neural networks. Initially, Finite Structured Acoustic Element Analysis (FE) was employed to identify the natural frequencies arising from exposure to high noise levels in the sandwich panel. Subsequently, the outputs of the FE model (comprising 224 configurations) were used to train an artificial neural network (ANN) model, using the Levenberg-Marquardt method. The developed ANN model was then utilized to assess the impact of various parameters, including changes in inner cell angles (−45 to +45°), alterations in material composition, variations in skin thickness, and total mass, on the transmission characteristics of the sandwich structure. The results generated by the ANN model closely align with those obtained from the numerical model, exhibiting a small error tolerance of 0.483 and a remarkable correlation coefficient of 0.99798. [ABSTRACT FROM AUTHOR] |
