Accelerated inverse design of broadband microperforated panel absorbers based on probabilistic generative model
The inverse design of acoustic metamaterials for broadband sound absorption remains challenging due to the complex coupling dynamics and the ill-posed nature of mapping target absorption spectra to geometric parameters. This study proposes a deterministic autoencoder-like framework that integrates r...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-09-01
|
| Series: | Results in Engineering |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2590123025027288 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The inverse design of acoustic metamaterials for broadband sound absorption remains challenging due to the complex coupling dynamics and the ill-posed nature of mapping target absorption spectra to geometric parameters. This study proposes a deterministic autoencoder-like framework that integrates response prediction and inverse design modeling, enabling highly accurate bidirectional mapping between geometric parameters and absorption spectra. To address the ill-posed nature of inverse problems, particularly when dealing with non-physical or user-defined target spectra, a probabilistic generative model—conditional Variational Autoencoders (cVAEs)—is employed. This model constructs a disentangled latent space, facilitating the generation of multiple feasible geometric solutions for a single user-defined spectras. This probabilistic generative model requires only four absorbers to achieve quasi-perfect sound absorption (absorption coefficient α≥0.9) across any user-defined frequency bands within 400∼1300 Hz. This significantly reduces the complexity and difficulty of the design process, as the target absorption spectra can be arbitrarily customized. Furthermore, the proposed method significantly accelerates the inverse design of microperforated panel, ensuring adaptability to both physically realizable and non-physical input spectra without compromising broadband noise absorption performance. |
|---|---|
| ISSN: | 2590-1230 |