Topological insulators are characterized by exhibiting an internal insulating state but a surface conductor state, which makes them advantageous for applications in novel devices. However, for most given acoustic topological metamaterials, the operating frequency is relatively fixed and the effect of temperature on their topological properties is rarely considered. Therefore, a temperature-tunable acoustic topological insulator is constructed in this paper. The quadruple degenerate Dirac cone formed at \(\Gamma\) point can be opened by adjusting the temperature, causing topological band inversion between the doubly degenerate dipolar and quadrupole states, and achieving topological phase transition. The evolution of its topological state with temperature is numerically investigated and a novel topological acoustic waveguide is constructed. The switching effect of temperature on the waveguide device is verified by numerical simulation and experiment. Non-contact active modulation of edge states in the structure is achieved by temperature-controlled topological phases, exhibiting acoustic switching effects. This study can provide corresponding references for the intelligent control of acoustic topology in noise, vibration, and other aspects.

Acoustic metamaterials have been widely applied in fields such as sound insulation and noise reduction due to their controllable band structures and unique abilities to manipulate low-frequency sound waves. However, there exists a highly nonlinear mapping relationship between their structural parameters and performance responses, which causes traditional design methods to face the problems of inefficiency and poor generalization. Therefore, this paper proposes a bidirectional modeling framework based on deep learning. We constructed a forward prediction network that integrates an attention mechanism, a multi-scale feature fusion, and a reverse design model that combines an improved autoencoder and cascaded neural network to efficiently model the dispersion performance of acoustic metamaterials. In the feedforward network, the improved forward prediction model shows superior performance compared to the traditional Convolutional Neural Network model and the model based only on the Convolutional Block Attention Module attention mechanism, with a prediction accuracy of 99.65%. It has better fitting ability and stability in the high-frequency part of the dispersion curve. In the inverse network part, compression of the high-dimensional dispersion curves by an improved autoencoder reduces the training time by about 13.5% without significant degradation of the inverse prediction accuracy. The proposed network model provides a more efficient method for the design of metamaterials.

The current deep convolution features based on retrieval methods cannot fully use the characteristics of the salient image regions. Also, they cannot effectively suppress the background noises, so it is a challenging task to retrieve objects in cluttered scenarios. To solve the problem, we propose a new image retrieval method that employs a novel feature aggregation approach with an attention mechanism and utilizes a combination of local and global features. The method first extracts global and local features of the input image and then selects keypoints from local features by using the attention mechanism. After that, the feature aggregation mechanism aggregates the keypoints to a compact vector representation according to the scores evaluated by the attention mechanism. The core of the aggregation mechanism is to allow features with high scores to participate in residual operations of all cluster centers. Finally, we get the improved image representation by fusing aggregated feature descriptor and global feature of the input image. To effectively evaluate the proposed method, we have carried out a series of experiments on large-scale image datasets and compared them with other state-of-the-art methods. Experiments show that this method greatly improves the precision of image retrieval and computational efficiency.

Acoustic metasurfaces play a key role in building acoustics, noise control, and acoustic cloaking by regulating the acoustic wave scattering characteristics through subwavelength structures. The design of diffusely reflecting metasurfaces aims to achieve a uniform distribution of a scattered field, which is essentially a high-dimensional nonconvex optimization problem that needs to balance the computational efficiency in the synergistic optimization of the spatial arrangement of cells and the angular response. In traditional methods, a heuristic algorithm is prone to local optimization, and it is difficult to balance the global search and local adjustment. And full-wave simulation is time consuming and seriously restricts the design efficiency. Therefore, the hybrid tornado-gradient descent optimization algorithm (VDGD) is proposed in this paper. It uses a two-stage collaborative optimization approach to refine the reflection angle distribution of acoustic metasurfaces, thereby enhancing the uniformity of the diffuse acoustic field. The Tornado Optimization Algorithm (TOA) was initially employed to introduce global perturbations to the randomly initialized design. Local optimization can be avoided by gradually decreasing the perturbation magnitude, which reduces the standard deviation of the sound field from about 5.81 dB to about 4.07 dB. Then, the gradient descent is used for local fine adjustment to further reduce the standard deviation to about 1.91 dB. Experimental results show that the VDGD algorithm outperforms the seven classical and up-to-date optimization algorithms in improving scattering uniformity. This method achieves an effective balance between global search and local fine tuning, providing an efficient and flexible optimization strategy for metasurface design, which can bring application support for intelligent acoustic devices and sound field regulation.

Linked Data, a new form of knowledge representation and publishing described by RDF, can provide more precise and comprehensible semantic structures. However, the current RDF Schema (RDFS) and SPARQL-based query strategy cannot fully express the semantics of RDF since they cannot unleash the implicit semantics between linked entities, so they cannot unleash the potential of Linked Data. To fill this gap, this chapter first defines a new semantic annotating and reasoning method which can extend more implicit semantics from different properties and proposes a novel general Semantically-Extended Scheme for Linked Data Sources to realize the semantic extension over the target Linked Data source. Moreover, in order to effectively return more information in the process of semantic data retrieval, we then design a new querying model which extends the SPARQL pattern. Lastly, experimental results show that our proposal has advantages over the initial Linked Data source and can return more valid results than some of the most representative similarity search methods.