Abstract

Acoustic metamaterials are artificial composite acoustic structures, whose properties mainly depend on their geometric structures and arrangement. Therefore, acoustic metamaterials have many novel physical properties that natural materials do not have. In the past two decades, people have used acoustic metamaterials to realize a lot of novel physical phenomena, such as negative refraction and anisotropic materials. And many acoustic devices, including acoustic lens and cloaks, have been constructed based on acoustic metamaterials, which provided a new material platform for the manipulation of sound propagation. However, the development of accurate and computationally efficient design and optimization approaches for the acoustic metamaterials is still in the early stages. On the one hand, we need to design the complex acoustic structures efficiently and accurately to meet the requirements, that is, the intelligent design of acoustic metamaterials. On the other hand, we also need to consider the adaptability of the structures with the signal processing algorithm during the design process to optimize the performance of the whole acoustic system, that is, the intelligent application of acoustic metamaterials. This paper is devoted to explore the technical problems existing in the intelligent design and application of acoustic metamaterials. The main contents of this paper include following aspects:

1. An optimization strategy based on the genetic algorithm is proposed to solve the problem that it is difficult to design complex acoustic structures manually. The proposed method can not only be combined with analytical algorithms, such as the layered rigorous coupled-wave analysis, to improve the sound absorption performance of the absorber in the target frequency range, but also be combined with numerical algorithms, such as finite element method, to optimize the stealth effect of the acoustic cloak. The performance superiority of the proposed method has been demonstrated through simulations and experiments.

2. In view of the high complexity of the optimization algorithms, we propose an efficient, flexible and universal acoustic structure design method to solve the inverse problem of the acoustic structures based on the deep learning and lumped-parameter technique (LPT). In our method, we analyze acoustic structures by LPT and develop connections between geometric parameters and equivalent electrical parameters. Then, we generate the datasets to train and test the deep learning model. To evaluate the effectiveness of the proposed model, a two-order Helmholtz resonator is designed to realize acoustic insulation at specific frequencies. Moreover, we design a composite structure with 9 two-order Helmholtz resonators using the proposed method to realize low-frequency broadband sound insulation.

3. Some extended applications of the deep learning model in the field of acoustic structure design are discussed. Considering that the corresponding relation between target acoustic functionality and acoustic structures is usually not a deterministic one-toone mapping, we combine the proposed model and the principal component analysis to search for multiple solutions of the inverse design problem and make further selections based on different properties. Moreover, the proposed model can provide a good initial condition for the optimization algorithm to avoid fall into local optimum and improve the optimization effect. Finally, considering that various acoustic structures can be analyzed by the LPT exactly in the low frequency range, the proposed approach has a strong versatility and scalability, which can be further extended to other acoustic structures.

4. We develope a metamaterial-based single-microphone listening system (MSLS), which can localize and separate multiple sound signals from an overlapping signal in three-dimensional(3D) space. A 3D metamaterial enclosure (ME) is used to provide monaural cues to the inversion task by coding sound signals as a function of the source direction, which is designed from the point of view of adaptation of the structures with the signal processing algorithm. During signal processing, a joint algorithm of variable sparsity principal component analysis and orthogonal matching pursuit (VSPCA-OMP) is used to solve the multi- source listening problem, which has the advantages of low computational complexity and good real-time performance. A lot of experiment results promise a wide range of potential applications for our proposed system, such as intelligent scene monitoring and robot audition.

In a word, this paper systematically makes an in-depth study on the intelligent design and application of acoustic metamaterials. Corresponding theoretical analysis and simulation calculations have been performed. According to the different physical characteristics and application requirements, several automatic design methods are proposed. The prototypes of new metamaterial devices are designed to realize novel and rich acoustic phenomena and functions. The proposed methods break through the current predicament of the structure design overly relying on the experience, which greatly save the labor cost and time cost. This study provides an important theoretical and technical basis for the design of various types of acoustic devices and acoustic systems, which further promotes the practical and industrial development of acoustic metamaterials.

Download paper here