Data-driven structural design of lightweight metal-fiberglass laminates for enhanced sound insulation

Abstract

Acoustic materials and composite structural design play an important role in mitigating low-frequency noise in transportation systems. This study proposes a data-driven modeling and structural design approach for lightweight metal-fiberglass laminated structures with improved sound insulation performance. Two laminated configurations were investigated: aluminum sheet-fiberglass-aluminum sheet (AFA) and galvanized sheet-fiberglass-aluminum honeycomb panel-fiberglass-galvanized sheet (GFAFG). A piecewise regression framework was developed to predict the sound transmission loss (STL) across different frequency ranges. The model parameters were estimated using the Levenberg-Marquardt (LM) nonlinear optimization algorithm based on experimental impedance tube measurements. The resulting prediction model achieved a mean absolute error of 1.15 dB. Structural optimization of the AFA configuration resulted in a surface density of 12.36 kg/m² and a fiberglass thickness of 4.079 cm, achieving an average STL of 19.74 dB. Orthogonal experimental analysis for the GFAFG structure revealed that fiberglass layers had the most significant influence on STL performance, followed by the second fiberglass layer and the aluminum honeycomb panel. The optimal configuration (GFAFG432) achieved an average STL of approximately 40 dB. The proposed data-driven prediction framework and the GFAFG laminated structure provide an effective solution for lightweight noise reduction structures in transportation systems.