Abstract:Ventilated acoustic resonator(VAR), a type of acoustic metamaterial, emerge as an alternative for sound attenuation in environments that require ventilation, owing to its excellent low-frequency attenuation performance and flexible shape adaptability. However, due to the non-linear acoustic responses of VARs, the VAR designs are generally obtained within a limited parametrized design space, and the design relies on the iteration of the numerical simulation which consumes a considerable amount of computational time and resources. This paper proposes an acoustic response-encoded variational autoencoder (AR-VAE), a novel variational autoencoder-based generative design model for the efficient and accurate inverse design of VAR even with non-parametrized designs. The AR-VAE matches the high-dimensional acoustic response with the VAR cross-section image in the dimension-reduced latent space, which enables the AR-VAE to generate various non-parametrized VAR cross-section images with the target acoustic response. AR-VAE generates non-parameterized VARs from target acoustic responses, which show a 25-fold reduction in mean squared error compared to conventional deep learning-based parameter searching methods while exhibiting lower average mean squared error and peak frequency variance. By combining the inverse-designed VARs by AR-VAE, multi-cavity VAR was devised for broadband and multitarget peak frequency attenuation. The proposed design method presents a new approach for structural inverse-design with a high-dimensional non-linear physical response.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: how to perform the inverse design of non - parametric ventilation acoustic resonators (VAR) efficiently and accurately in order to achieve specific acoustic responses. Specifically, traditional VAR design methods rely on parametric design spaces and numerical simulation iterations, which consume a large amount of computational resources and time, and are difficult to handle complex geometries and high - dimensional non - linear physical responses. Therefore, the author proposes a new generative model based on variational auto - encoders (VAE) - the acoustic - response - encoding variational auto - encoder (AR - VAE) for the inverse design of non - parametric VAR. ### Specific background of the problem 1. **Limitations of traditional methods**: - **Parametric design space**: Traditional methods are usually limited to parametric design spaces, which limit the degree of design freedom. - **Numerical simulation iteration**: Methods relying on numerical simulation iteration consume a large amount of computational resources and time. - **Complex geometries**: As the geometric complexity increases, inverse design based on human intuition or analytical methods becomes increasingly difficult. 2. **Research objectives**: - **Efficient inverse design**: Develop a design method that can generate non - parametric VARs with specific acoustic responses efficiently and accurately. - **High - degree - of - freedom design space**: Explore a broader design space without relying on parametric design. - **Error reduction**: Significantly reduce the mean - square error (MSE) and peak - frequency deviation compared to traditional deep - learning methods. ### Solution The AR - VAE model proposed by the author solves the above problems in the following ways: 1. **Acoustic - response encoding**: Encode high - dimensional acoustic responses into a low - dimensional latent space, enabling the model to generate VAR cross - sectional images similar to the target acoustic responses. 2. **Non - parametric design**: The generated VAR cross - sectional images are non - parametric, allowing for more flexible and diverse geometries. 3. **Efficient generation**: Compared to traditional methods, AR - VAE can generate multiple candidate designs in a shorter time and select the optimal solution from them. ### Main contributions 1. **25 - fold reduction in mean - square error**: Compared to traditional deep - learning methods, the VAR generated by AR - VAE has a 25 - fold reduction in the mean - square error in the target acoustic response. 2. **Lower average mean - square error and peak - frequency deviation**: Non - parametric inverse - designed VARs exhibit lower average mean - square error and peak - frequency deviation. 3. **Multi - cavity VAR design**: By combining multiple non - parametric inverse - designed single VARs, broadband multi - peak - frequency acoustic attenuation is achieved. In summary, this paper aims to solve the problems of high computational resource consumption, limited design space, and difficulty in handling complex geometries in traditional VAR inverse - design methods by introducing the AR - VAE model, thereby providing a new and efficient solution for the inverse design of acoustic structures.

Inverse design of Non-parameterized Ventilated Acoustic Resonator via Variational Autoencoder with Acoustic Response-encoded Latent Space

Adversarial-Network Regularized Inverse Design of Frequency-Selective Surface with Frequency-Temporal Deep Learning

Ultrasparse and Omnidirectional Acoustic Ventilated Meta-Barrier

Inverse design of acoustic metamaterials based on machine learning using a Gauss–Bayesian model

Acoustic Metamaterial Inverse Design Based on Machine Learning Using Gauss–Bayesian Models

CVAE-Based Inverse Design of Two-Dimensional Honeycomb Pentamode Metastructure for Acoustic Cloaking

Inverse Design Framework With Invertible Neural Networks for Passive Vibration Suppression in Phononic Structures

On-demand Inverse Design of Acoustic Metamaterials Using Probabilistic Generation Network

Learning to inversely design acoustic metamaterials for enhanced performance

Machine learning assisted design of acoustic metamaterials with broadband sound-absorbing and superior mechanical performance

Inverse design of a Helmholtz resonator based low-frequency acoustic absorber using deep neural network

Design of Phononic Bandgap Metamaterials Based on Gaussian Mixture Beta Variational Autoencoder and Iterative Model Updating

Sound Design Strategies for Latent Audio Space Explorations Using Deep Learning Architectures

Generative inverse design of multimodal resonant structures for locally resonant metamaterials

Parametric modeling and deep learning-based forward and inverse design for acoustic metamaterial plates

A Gaussian Mixture Variational Autoencoder-Based Approach for Designing Phononic Bandgap Metamaterials

Reverse Design of Absorption Performance for Typical Underwater Acoustic Coatings Based on Neural Network

Inverse design of laminated plate-type acoustic metamaterials for sound insulation based on deep learning

Reverse design and application of phononic crystals based on deep learning

Generative deep learning model for broadband acoustic metamaterial design

A Vector Quantized Variational Autoencoder (VQ-VAE) Autoregressive Neural $F_0$ Model for Statistical Parametric Speech Synthesis