AI-Driven Optimization of Acoustic Metamaterials for Low-Frequency Noise Attenuation in Aerospace Applications

The need for advanced noise control solutions in aerospace applications has encouraged research in the design and optimization of acoustic metamaterials, engineered configurations capable of addressing low-frequency noise where traditional materials often fail. This paper presents an AI-driven methodology employing deep neural networks (DNN) within an autoencoder architecture to design and optimize acoustic metamaterials for improved noise attenuation. The autoencoder framework leverages an encoder to extract latent features from high-dimensional input data and a decoder to predict five critical geometric parameters: neck diameter, neck thickness, slit diameter, slit thickness, and the number of periodic unit cells (PUC). These parameters directly influence the ability of the acoustic metamaterial to absorb sound, particularly at resonance frequencies. This approach achieves reliable and efficient designs capable of absorbing at least 50% of sound energy at target frequencies, addressing the significant challenges posed by low-frequency noise in aerospace environments. By combining advanced machine learning techniques with acoustic modeling, the developed framework offers a scalable, data-driven solution for optimizing metamaterial configurations. This work highlights the potential of integrating deep learning with acoustic design to create innovative noise control solutions, advancing the field of aerospace acoustics and paving the way for future research into AI-optimized metamaterials.