Double-negative acoustic metamaterial structure based on complementary medium model and its inverse design method

Abstract

The application provides a double-negative acoustic metamaterial structure based on a complementary medium model and a reverse design method thereof, comprising: a plurality of axially stacked labyrinth layers, and a partition plate arranged between adjacent layers; wherein the plurality of axially stacked labyrinth layers is a multilayer discretization structure, and the multilayer discretization structure forms a stepped approximation; the stepped approximation is to provide a segmented constant equivalent bulk modulus in a respective corresponding radial interval of a plurality of discrete labyrinth layers to approximate a continuous radial distribution of a target bulk modulus; the labyrinth layer adopts a periodic coiled sound channel structure; and the coiled sound channel structure realizes negative equivalent mass density and negative equivalent bulk modulus by lengthening a sound propagation path. The acoustic metamaterial provided by the application can realize significant scattering enhancement and suppression of dipole sound radiation in a low-frequency range, has a good control effect on unmanned aerial vehicle propeller noise, and has the advantages of compact structure, adjustable parameters, suitability for engineering design and batch manufacturing, etc.

Description

Technical Field

[0001] This invention relates to the field of acoustic metamaterials and noise control technology, specifically to a double negative acoustic metamaterial structure based on a complementary medium model and its reverse design method, and more specifically to a multilayer acoustic metamaterial suitable for the dipole noise control of UAV propellers and its intelligent optimization design method and system. Background Technology

[0002] With the widespread application of unmanned aerial vehicles (UAVs) in logistics, low-altitude inspection, and urban air traffic, aerodynamic noise has gradually become a significant factor limiting their further promotion. Among these noises, the blade passage frequency (BPF) and its harmonic noise generated by propeller rotation typically exhibit a distinct dipole radiation characteristic, characterized by prominent low frequencies, long propagation distances, and difficulty in being effectively suppressed by traditional passive noise reduction methods.

[0003] Traditional sound-absorbing materials mainly rely on viscous dissipation to achieve noise attenuation, which often requires a large structural thickness in the low-frequency range, making it difficult to meet the strict weight and space constraints of UAV systems.

[0004] In recent years, acoustic metamaterials have gradually become an important research direction for low-frequency noise control due to their ability to achieve unconventional equivalent acoustic parameters through subwavelength structures. Among them, the complementary media model based on transform acoustics theory can realize phenomena such as scattering boundary expansion, superscattering, and radiation suppression through coordinate transformation, providing new ideas for source noise control.

[0005] However, existing acoustic metamaterial designs still have the following problems: 1. Double negative acoustic parameters are difficult to achieve directly from natural materials; 2. Ideal continuous parameter distributions cannot be directly manufactured; 3. The high dimensionality of the parameter space in multi-layered structures leads to low efficiency of traditional parameter scanning and heuristic optimization. 4. Geometric variables contain both continuous and discrete variables, and the inverse mapping exhibits obvious non-uniqueness; 5. Multi-layered structures have strict geometric continuity constraints, which can easily lead to unmanufacturable design results.

[0006] Therefore, there is an urgent need to propose an acoustic metamaterial structure and design method that can achieve the target parameter distribution of complementary media and efficiently complete reverse design under complex geometric constraints. Summary of the Invention

[0007] To address the shortcomings of existing technologies, the purpose of this invention is to provide a double negative acoustic metamaterial structure based on a complementary medium model and its reverse design method.

[0008] According to the present invention, a double negative acoustic metamaterial structure based on a complementary medium model includes: multiple axially stacked labyrinth layers, with partitions between adjacent labyrinth layers; wherein the labyrinth layers are multi-layer discretized structures, and the multi-layer discretized structures form a stepped approximation; the stepped approximation is to provide piecewise constant equivalent bulk modulus in the corresponding radial intervals of the multiple discrete labyrinth layers, so as to approximate the continuous radial distribution of the target bulk modulus; The labyrinth layer adopts a periodic curled acoustic channel structure; the curled acoustic channel structure achieves negative equivalent mass density and negative equivalent bulk modulus by extending the sound propagation path.

[0009] Preferably, the maze layer includes: an outer ring boundary, multiple circumferentially distributed curled channels, a sector segmentation structure, and a central region; Wherein, the outer ring boundary is located at the outermost radial direction of the maze layer; the central region is located at the center of the maze layer; the sector segmentation structure is radially connected between the outer ring boundary and the central region, and divides the annular region between them into multiple sectors; the curled audio channel is set in each of the sectors.

[0010] Preferably, the circumferential sector corresponds to the central angle. for:

[0011] Inner diameter satisfies:

[0012] in, This refers to the number of circumferential sectors. R is the outer radius of the outer ring boundary of the maze layer; i The inner radius of the central region; This refers to the width of the audio channel; Number of audio channels; The acoustic tract wall is thick.

[0013] Preferably, the plurality of axially stacked labyrinth layers satisfy the target bulk modulus distribution radially:

[0014] in, R is the radial coordinate from the center of the maze layer; κ(r) is the target equivalent bulk modulus at that radial coordinate; o The outer radius of the outer ring boundary of the maze layer is 2R; the outer diameter of the maze layer is 2R. o .

[0015] The present invention provides a reverse design method for a double negative acoustic metamaterial structure based on a complementary medium model, comprising: Step S1: Establish a complementary medium model based on transform acoustics theory and derive the distribution of the target equivalent parameters; Step S2: Establish the mapping relationship between geometric parameters and equivalent parameters using a multilayer perceptron; Step S3: Use the beam search algorithm to generate some multi-layer candidate structures along the layer sequence from the outside to the inside; for the first layer, generate multiple single-layer candidate structures within the range of preset geometric parameters; for the i-th layer, based on the partially multi-layer candidate structures retained in the (i-1)-th layer, generate the i-th layer single-layer candidate structure under the conditions of satisfying the inner diameter, interlayer continuity and preset manufacturability constraints of the (i-1)-th layer, and combine it with its corresponding partially multi-layer candidate structures. Step S4: Use a multilayer perceptron to predict the equivalent parameters of each single-layer candidate structure, and score the expanded partial multilayer candidate structures based on target equivalent parameter matching, interlayer continuity constraints, terminal inner diameter constraints, and thickness uniformity constraints; retain the top K partial multilayer candidate structures with the highest scores as the current layer candidate retention set, and use the candidate retention set as the constraint basis for generating the next single-layer candidate structure, until the preset number of layers or target inner diameter is reached.

[0016] Preferably, step S1 includes: Target equivalent mass density satisfy:

[0017] The target bulk modulus satisfies: ; in, R is the radial coordinate from the center of the maze layer; κ(r) is the target equivalent bulk modulus at that radial coordinate; o is the outer radius of the outer ring boundary of the maze layer.

[0018] Preferably, step S2 includes: parametrically sampling the outer diameter, wall thickness, channel width, number of channel turns, and number of sectors; performing finite element simulation on each set of geometric parameters to obtain the corresponding equivalent mass density and equivalent bulk modulus, forming training sample pairs; training a multilayer perceptron with the geometric parameters as input and the equivalent acoustic parameters as output, and obtaining a forward surrogate model for predicting the equivalent parameters of candidate structures after training.

[0019] Preferably, in step S4, the extended partial multilayer candidate structures are scored based on target equivalent parameter matching, interlayer continuity constraints, terminal inner diameter constraints, and thickness uniformity constraints, including:

[0020] in, A comprehensive score for a complete multi-layered structure; The local score for the candidate structure at layer i; The penalty incurred when the inner diameter of the terminal deviates from the target inner diameter; and These are the weighting coefficients for the corresponding penalty items; The penalty for uneven distribution of thickness in each layer.

[0021] Preferably, the step of retaining the top K candidate structures based on the score and expanding the next layer of candidate structures until a preset number of layers or a target inner diameter is reached includes: If the overall score of the optimal candidate structure does not reach the preset threshold when the preset number of layers or target inner diameter is reached, the beam width K is increased and / or the sampling range of the candidate geometric parameters is adjusted before re-executing candidate generation, prediction and scoring; when the preset maximum number of iterations is reached but the threshold is still not met, the current optimal candidate structure is output and its unmet constraints are recorded for further adjustment of target parameters or creation of constraints.

[0022] The present invention provides a reverse design system for a double negative acoustic metamaterial structure based on a complementary medium model, comprising: Module M1: Establishes a complementary medium model based on transform acoustics theory and derives the target equivalent parameter distribution; Module M2: Employs a multilayer perceptron to establish a mapping relationship between geometric parameters and equivalent parameters; Module M3: A bundle search algorithm is used to generate some multi-layer candidate structures along the layer sequence from the outside to the inside. For the first layer, multiple single-layer candidate structures are generated within the range of preset geometric parameters. For the i-th layer, based on the partial multi-layer candidate structures retained in the (i-1)-th layer, a single-layer candidate structure of the i-th layer is generated under the conditions of satisfying the inner diameter, interlayer continuity and preset manufacturability constraints of the (i-1)-th layer, and combined with its corresponding partial multi-layer candidate structures. Module M4: Utilizes a multilayer perceptron to predict the equivalent parameters of each single-layer candidate structure, and scores the expanded partial multilayer candidate structures based on target equivalent parameter matching, interlayer continuity constraints, terminal inner diameter constraints, and thickness uniformity constraints; retains the top K partial multilayer candidate structures with the highest scores as the current layer candidate retention set, and uses the candidate retention set as the constraint basis for generating the next single-layer candidate structure, until the preset number of layers or target inner diameter is reached.

[0023] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention achieves low-frequency sound field modulation based on a complementary medium model, which can effectively suppress dipole sound radiation; 2. This invention approximates the continuous target parameter distribution through a multi-layer discretization structure, thereby improving engineering feasibility; 3. This invention employs a double negative acoustic metamaterial structure, which can achieve scattering enhancement effects that are difficult to obtain with traditional materials; 4. This invention uses an MLP proxy model to replace repetitive finite element simulation, significantly reducing the amount of optimization computation; 5. This invention employs a beam search algorithm to achieve layer-by-layer optimization under complex geometric constraints; 6. This invention controls both local parameter matching and overall structural regularity simultaneously through multiple penalty functions; 7. This invention is applicable to low-frequency noise control of UAV propellers and has the advantages of compact structure, strong scalability, and suitability for mass production; 8. This invention enables effective control of dipole-type low-frequency acoustic radiation and solves the problems of low efficiency, complex geometric constraints, and difficulty in engineering implementation of existing metamaterial reverse design. Attached Figure Description

[0024] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 A schematic diagram of the overall structure of a multilayer double-negative acoustic metamaterial; Figure 2 A schematic diagram of the maze's layer structure and geometric parameters; Figure 3 The near-field sound pressure distribution and far-field radiation pattern are shown. Figure 4 The sound pressure field distribution is shown with and without the acoustic metamaterial structure installed. Figure labels: 1-Multi-layer double negative acoustic metamaterial; 11-Maze layer; 12-Separator layer; 111-Outer ring boundary; 112-Curled channel; 113-Sector segmentation structure; 114-Central region. Detailed Implementation

[0025] The present invention will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present invention, but do not limit the invention in any way. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the present invention. These all fall within the protection scope of the present invention.

[0026] Example 1 According to the present invention, a double negative acoustic metamaterial structure based on a complementary medium model includes: multiple axially stacked labyrinth layers 11, with partition layers 12 disposed between adjacent labyrinth layers; wherein, the labyrinth layer 11 is a multi-layer discretized structure, and the multi-layer discretized structure forms a stepped approximation; the stepped approximation is to provide piecewise constant equivalent bulk modulus in the corresponding radial intervals of the multiple discrete labyrinth layers, so as to approximate the continuous radial distribution of the target bulk modulus; That is, the maze layers 11 together constitute the multi-layer structure; each maze layer 11 is an independent structural unit, and adjacent maze layers are separated by partition layers 12, thereby forming a multi-layer discretized structure stacked along the axial direction.

[0027] The labyrinth layer 11 adopts a periodic curled acoustic channel structure; the curled acoustic channel structure achieves negative equivalent mass density and negative equivalent bulk modulus by extending the sound propagation path.

[0028] The maze layer 11 includes: an outer ring boundary 111, multiple circumferentially distributed curled channels 112, a sector segmentation structure 113, and a central region 114. Wherein, the outer ring boundary 111 is located on the outermost radial side of the maze layer 11; the central region 114 is located at the center of the maze layer 11; the sector segmentation structure 113 is radially connected between the outer ring boundary 111 and the central region 114, and divides the annular region between them into multiple sectors; the curled audio channel 112 is disposed in each of the sectors.

[0029] Specifically, the central angle corresponding to the circumferential sector is:

[0030] Inner diameter satisfies:

[0031] in, α represents the number of circumferential sectors; α is the central angle corresponding to a single sector. R is the outer radius of the outer ring boundary of the maze layer; i The inner radius of the central region; This refers to the width of the audio channel; The number of channel layers arranged radially within a single sector; The acoustic tract wall is thick.

[0032] Specifically, the plurality of axially stacked labyrinth layers 11 satisfy the target bulk modulus distribution radially:

[0033] in, R is the radial coordinate from the center of the maze layer; κ(r) is the target equivalent bulk modulus at that radial coordinate; o The outer radius of the outer ring boundary of the maze layer is 2R; the outer diameter of the maze layer is 2R. o .

[0034] The multi-layer structure refers to multiple labyrinth layers stacked along the axial direction. Within each radial interval, the continuous values ​​of the target bulk modulus κ(r) are replaced by the equivalent bulk modulus of the corresponding labyrinth layer, thereby performing a piecewise constant approximation of the continuous radial distribution. The term "stepped approximation" means that the equivalent bulk modulus of adjacent radial intervals changes stepwise according to the intervals, and the deviation from the target distribution is not greater than a preset error threshold.

[0035] In this embodiment, the maze layer adopts a periodic curled channel structure to achieve double negative equivalent parameters.

[0036] The structure is used to control dipole acoustic radiation and is suitable for suppressing drone propeller noise.

[0037] The acoustic metamaterial proposed in this invention can achieve significant scattering enhancement and dipole acoustic radiation suppression in the low-frequency range, and has a good control effect on the propeller noise of UAVs. It also has the advantages of compact structure, adjustable parameters, and suitability for engineering design and mass production.

[0038] Example 2 The reverse design method for a double negative acoustic metamaterial structure based on a complementary medium model provided by the present invention includes the following steps: "determination of target equivalent parameters—establishment of forward surrogate model—constrained layer-by-layer bundle search—scoring and screening". Step 1: Determine the distribution of the target equivalent parameters; Specifically, step 1 includes: establishing a complementary medium model based on transform acoustics theory and deriving the target equivalent parameter distribution.

[0039] The target equivalent mass density satisfies:

[0040] The target bulk modulus satisfies:

[0041] Step 2: Train the forward surrogate model to obtain geometric parameters to equivalent parameters; Specifically, step 2 includes: Specifically, the outer diameter, wall thickness, channel width, number of channel turns, and number of sectors are parameterized and sampled; finite element simulation is performed on each set of geometric parameters to obtain the corresponding equivalent mass density and equivalent bulk modulus, forming training sample pairs; a multilayer perceptron is trained with the geometric parameters as input and the equivalent acoustic parameters as output, and a forward surrogate model for quickly predicting the equivalent parameters of candidate structures is obtained after training.

[0042] Step 3: Generate the next layer of single-layer candidate structures under the constraints of the multi-layer candidate structures retained in the previous layer; Specifically, step 3 includes: A bundle search algorithm is used to generate some multi-layer candidate structures along the layer sequence from the outside to the inside. For the first layer, multiple single-layer candidate structures are generated within the range of preset geometric parameters. For the i-th layer, based on the partial multi-layer candidate structures retained in the (i-1)-th layer, a single-layer candidate structure of the i-th layer is generated under the conditions of satisfying the inner diameter, interlayer continuity and preset manufacturability constraints of the (i-1)-th layer, and combined with its corresponding partial multi-layer candidate structures. Step 4: Predict, score, and filter the bundles of the expanded multi-layer candidate structures, and use the retained results as the constraint basis for generating the next layer of candidate structures.

[0043] Specifically, step 4 includes: The equivalent mass density and equivalent bulk modulus of the i-th single-layer candidate structure are predicted using a forward surrogate model. Scores for some multi-layer candidate structures are calculated based on target parameter matching, inter-layer continuity constraints, geometric parameter range constraints, and overall regularity. Among all the multi-layer candidate structures in the i-th layer, only the top K with the highest scores are retained as the i-th layer candidate retention set. Each multi-layer candidate structure in the candidate retention set serves as the constraint basis for generating the (i+1)-th single-layer candidate structure. Steps 3 and 4 are executed iteratively in a "generation-prediction-scoring-retention" manner until the preset number of layers or the target inner diameter is reached.

[0044] Step 4 is used to quantify the target parameter matching degree, manufacturability constraints and overall regularity into a comprehensive score, so that the candidate structures generated in step 3 have a comparable order of merit; this comprehensive score serves as the basis for retaining, eliminating and expanding candidate structures in the bundle search.

[0045] More specifically,

[0046] in, A comprehensive score for a complete multi-layered structure; The local score for the candidate structure at layer i; The penalty incurred when the inner diameter of the terminal deviates from the target inner diameter; The penalty for uneven distribution of layer thickness; and These are the weight coefficients for the corresponding penalty terms. Maximizing... Candidate structures that balance target parameter matching, inner diameter constraints, and thickness uniformity are selected.

[0047] As one embodiment of the present invention, only the top K candidate structures with the highest scores are retained during the beam search process to reduce the computational complexity caused by combinatorial explosion.

[0048] If the overall score of the optimal candidate structure does not reach the preset threshold when the preset number of layers or target inner diameter is reached, the beam width K is increased and / or the sampling range of the candidate geometric parameters is adjusted before re-executing candidate generation, prediction and scoring; when the preset maximum number of iterations is reached but the threshold is still not met, the current optimal candidate structure is output and its unmet constraints are recorded for further adjustment of target parameters or creation of constraints.

[0049] The present invention also provides a reverse design system for a double negative acoustic metamaterial structure based on a complementary medium model. The reverse design system for the double negative acoustic metamaterial structure based on a complementary medium model can be implemented by executing the process steps of the reverse design method for the double negative acoustic metamaterial structure based on a complementary medium model. That is, those skilled in the art can understand the reverse design method for the double negative acoustic metamaterial structure based on a complementary medium model as a preferred embodiment of the reverse design system for the double negative acoustic metamaterial structure based on a complementary medium model.

[0050] Example 3 Example 3 is a preferred example of Example 1. This embodiment provides a multilayer double negative acoustic metamaterial structure, including: a labyrinth layer and a partition layer; In this embodiment, the double negative acoustic metamaterial structure based on the complementary medium model includes three labyrinth layers, with the partition layer located between adjacent labyrinth layers.

[0051] like Figure 1 The diagram shows a three-layer double negative acoustic metamaterial structure, including a labyrinth layer 11 and a partition layer 12, both made of PLA plastic. The thickness of a single labyrinth layer is 19 mm, the thickness of the partition layer is 1.5 mm, and the total thickness is 60 mm.

[0052] Example 4 Example 4 is a preferred example of Example 1. This embodiment provides a method for geometric parameterization of a double negative acoustic metamaterial layer, including: Each layer of double negative acoustic metamaterial comprises several sector structures arranged periodically along the circumference, with a coiled acoustic channel within each sector. The coiled acoustic channel is used to extend the sound propagation path, causing the sound wave to have an equivalent phase delay in the subwavelength scale structure, thereby achieving negative equivalent mass density and negative equivalent bulk modulus.

[0053] like Figure 2 The diagram shows the structure and geometric parameters of the maze layers. The structure of each layer is determined by the following five geometric parameters: the outer radius R of the outer ring boundary of the maze layer. o ; Channel wall thickness Channel width w; Number of channel layers M; Number of sectors N.

[0054] Wherein, the central angle α of the sector satisfies:

[0055] The inner diameter R of each layer i Determined by the following formula:

[0056] Among them, R i The inner radius of the central region; In one specific implementation, the wall thickness t can take discrete values, such as 0.001m, 0.0012m or 0.0014m; the number of channel turns M can take integer values, such as 1 to 8; the number of sectors N can be selected according to the requirements of approximate isotropy, processing difficulty and acoustic performance requirements, such as several integers from 4 to 12; the channel width w can be optimized as a continuous variable.

[0057] In another specific embodiment, a spacer layer is provided between the double negative acoustic metamaterial layers. The spacer layer is used to maintain the axial distance between adjacent double negative acoustic metamaterial layers and to ensure the stacking stability of the multilayer structure in three-dimensional space. The thickness of the spacer layer can be determined according to the structural strength, processing method, and acoustic coupling effect.

[0058] This embodiment describes the double negative acoustic metamaterial layer using a five-parameter approach, enabling the complex coiled channel structure to be uniformly incorporated into the reverse design and optimization process, while also facilitating subsequent parameter search, finite element modeling, and engineering manufacturing.

[0059] Example 5 Example 5 is a preferred example of Example 2. The present invention provides a method for constructing target parameters of complementary media based on transformation acoustics theory, which is used to determine the target equivalent parameters to be approximated for a multilayer double negative acoustic metamaterial structure.

[0060] First, based on the invariance of the acoustic wave equation under coordinate transformation, a radial coordinate transformation relationship is constructed. By folding the coordinate transformation, a portion of the virtual space is mapped to a finite region in the physical space, causing the equivalent acoustic parameters within that region to exhibit a sign reversal, thereby forming a complementary medium effect.

[0061] In this embodiment, the target equivalent mass density is set as follows:

[0062] The target equivalent bulk modulus is set as a function of radial position: ; in, R is the radial coordinate from the center of the maze layer; κ(r) is the target equivalent bulk modulus at that radial coordinate; o It is the outer radius of the outer ring boundary of the maze layer.

[0063] Since the continuous distribution of bulk modulus is difficult to achieve directly from actual materials, this embodiment employs a multi-layer discretization method to approximate the continuous radial distribution as several piecewise constant values. Each labyrinth layer corresponds to a radial interval, and its equivalent mass density and equivalent bulk modulus are adjusted through its own geometric parameters, so that the multi-layer structure as a whole approximates the parameter distribution of the target complementary medium.

[0064] In this embodiment, the multi-layer structure is designed sequentially from the outside in. The outer radius of each layer is determined by the inner radius of the layer above it, or, within an allowable error range, matches the inner radius of the layer above it, thereby ensuring geometric continuity between adjacent layers. The target bulk modulus for each layer can be expressed as an objective function within the radial interval of that layer. The average value.

[0065] This embodiment transforms the continuous parameter distribution of an ideal complementary medium into a multilayer discrete structure design objective, thereby converting a difficult-to-manufacture continuous double negative medium into an engineering structure that can be realized through acoustic metamaterials.

[0066] Example 6 Example 6 is a preferred example of Example 2. This embodiment provides a method for constructing a forward surrogate model to predict the equivalent parameters of a double negative acoustic metamaterial layer, including: First, a parametric finite element model of the double negative acoustic metamaterial layer is established. This is achieved by changing the outer diameter R. o Multiple candidate structures were generated by considering factors such as wall thickness t, channel width w, number of channel turns M, and number of sectors N. The acoustic response of each candidate structure was then calculated using finite element simulation, and its equivalent mass density ρ and equivalent bulk modulus κ were obtained based on the equivalent parameter inversion method.

[0067] In this embodiment, a multilayer perceptron (MLP) is used as the forward proxy model. The input to this model is a five-dimensional geometric parameter vector: X=[R o ,t,w,M,N]; The output is a two-dimensional equivalent acoustic parameter vector: Y=[ρ,κ] In one specific implementation, a multilayer perceptron includes an input layer, several hidden layers, and an output layer. The hidden layers may be fully connected layers and may use the ReLU activation function. The output layer is used to output the equivalent mass density and equivalent bulk modulus of the corresponding structure.

[0068] In another specific implementation, the training data is generated by COMSOL Multiphysics simulation and divided into training and test sets in a certain ratio, such as 8:2. After training, the forward proxy model is used to quickly predict the equivalent acoustic parameters corresponding to candidate geometries during reverse engineering, thereby replacing a large number of repetitive finite element simulations.

[0069] In this embodiment, the forward surrogate model can significantly reduce the cost of candidate structure evaluation, enabling the subsequent beam search algorithm to perform efficient searches within a larger parameter space.

[0070] Example 7 Example 7 is a preferred example of Example 2. The present invention provides a method for reverse design of multilayer double negative acoustic metamaterial structures based on a beam search algorithm, comprising: First, the target equivalent parameter distribution is determined according to Example 5; multiple single-layer candidate structures of the first layer are generated within the preset geometric parameter range, and the equivalent mass density and equivalent bulk modulus of each single-layer candidate structure are predicted using the MLP forward surrogate model described in Example 6.

[0071] In this embodiment, the search process starts from the outermost layer and constructs a multi-layer structure layer by layer radially inward. For the first layer, multiple single-layer candidate structures are generated to form corresponding partial multi-layer candidate structures. For the i-th layer, based on each partial multi-layer candidate structure in the (i-1)-th layer candidate set, a single-layer candidate structure for the i-th layer is generated under the conditions of satisfying the inner diameter, interlayer continuity, and preset manufacturability constraints of the (i-1)-th layer. After combining the generated single-layer candidate structure for the i-th layer with its corresponding partial multi-layer candidate structure, the equivalent mass density and equivalent bulk modulus of the layer are predicted using an MLP forward surrogate model, and the score of the expanded partial multi-layer candidate structure is calculated based on a scoring function. After each layer expansion is completed, only the top K partial multi-layer candidate structures with the highest scores are retained as the candidate set for the next layer search, where K is the bundle width. In this way, while avoiding combinatorial explosion caused by exhaustive search, better candidate design paths are retained.

[0072] In one specific implementation, the scoring of a single-layer candidate structure includes the following penalty: Density matching penalty is used to constrain the equivalent mass density to be close to the target value ρ=-1; The bulk modulus deviation penalty is used to constrain the equivalent bulk modulus to be close to the average value of the target bulk modulus within the current radial interval. Bulk modulus range penalty is used to constrain the bulk modulus of the candidate structure to be within the target allowable range corresponding to the radial interval of that layer; Monotonicity penalty is used to constrain the bulk modulus variation trend of multilayer structures to be consistent with the variation trend of the objective function; Interlayer continuity penalty is used to constrain the difference between the outer diameter of the i-th single-layer candidate structure and the inner diameter of the (i-1)-th single-layer candidate structure in the multi-layer candidate structure of its part; Terminal inner diameter penalty is used to constrain the final inner diameter to be close to the preset target inner diameter; Global thickness uniformity penalty is used to avoid excessive uneven distribution of thickness among layers.

[0073] In one specific implementation, the overall scoring function can be expressed as:

[0074] in, This represents the local score of the i-th single-layer candidate structure relative to the target parameter, and together with the scores of the previous layers, it constitutes the cumulative score of a partial multi-layer candidate structure. Indicates the penalty for the inner diameter of the terminal. This represents a penalty for global thickness uniformity. and These are the corresponding weighting coefficients.

[0075] In one specific implementation, the beam search algorithm includes the following steps: Step 1: Generate multiple single-layer candidate structures for the first layer within the preset geometric parameter range, and use each single-layer candidate structure as a partial multi-layer candidate structure for the first layer; Step 2: Use the MLP forward surrogate model to predict the equivalent mass density and equivalent bulk modulus of each single-layer candidate structure; Step 3: Calculate the scores of the expanded partial multilayer candidate structures based on the distribution of target equivalent parameters, interlayer continuity constraints, and geometric parameter range constraints; Step 4: Retain the top K multi-layer candidate structures with the highest scores as the candidate retention set for the current layer; Step 5: Based on each part of the multi-layer candidate structure in the current layer candidate retention set, and under the conditions of satisfying the innermost inner diameter, inter-layer continuity and preset manufacturability constraints of the multi-layer candidate structure, generate the next layer of single-layer candidate structure and combine it with it. Step 6: Repeat steps 2 to 5 for each single-layer candidate structure generated in the next layer until the preset number of layers or the target inner diameter is reached; Step 7: Perform terminal scoring on the complete multi-layer structure; Step 8: Output one or more sets of multi-layer structure parameters with the best scores.

[0076] This embodiment transforms the complex problem of reverse design of acoustic metamaterials into a layer-by-layer decision-making problem, enabling the algorithm to explicitly handle continuous variables, discrete variables, and inter-layer geometric constraints, thus avoiding the problem of unmanufacturable structures that are easily generated by traditional end-to-end reverse neural networks.

[0077] Example 8 A method for numerically verifying a designed acoustic metamaterial structure according to the present invention includes: First, based on the multilayer structure parameters output in Example 7, a three-dimensional finite element model is established in COMSOL Multiphysics. The model includes an air domain, an acoustic metamaterial structure, and a perfectly matched layer for absorbing boundary reflections. The acoustic calculations can utilize the pressure acoustics module and can be combined with the solid mechanics module to consider acoustic-structure interaction effects.

[0078] In one specific embodiment, a plane sound wave is incident from one side of the structure and propagates horizontally. Four comparative models are established: the multilayer acoustic metamaterial structure designed in this invention; an ideal complementary medium model; a rigid cylindrical obstacle with the same outer diameter as the actual structure; and a rigid cylindrical obstacle with the same equivalent scattering radius.

[0079] By simulating the above four models, their near-field sound pressure distribution and far-field radiation pattern were obtained, such as... Figure 3 As shown.

[0080] Simulation results show that the multilayer acoustic metamaterial structure designed in this invention closely resembles the ideal complementary medium model and the equivalent scattering radius rigid barrier in terms of near-field sound pressure distribution and far-field radiation characteristics, while significantly outperforming the scattering effect of a rigid barrier with the same outer diameter. This indicates that the structure described in this invention can simulate a larger scattering boundary within a limited physical size, thereby achieving the sound field modulation function expected by the complementary medium model.

[0081] In another specific implementation, a dipole sound source is used to simulate the propeller noise of a UAV. The frequency of the dipole sound source can be set to the target propeller frequency or its corresponding characteristic frequency, such as 800Hz. The sound pressure field distribution is calculated separately for the cases without and with the acoustic metamaterial structure installed. Figure 4 As shown.

[0082] Simulation results show that without the acoustic metamaterial structure, the dipole sound source exhibits obvious symmetrical radiation characteristics; after installing the acoustic metamaterial structure described in this invention, the sound field distribution is significantly compressed and redistributed, and the far-field radiation intensity is reduced, indicating that this invention can effectively suppress dipole sound radiation.

[0083] This embodiment demonstrates, through three types of numerical verification—near-field, far-field, and dipole source—that the acoustic metamaterial structure designed in this invention can approximate the acoustic behavior of an ideal complementary medium model and has the potential for low-frequency noise control of UAV propellers.

[0084] In the description of this application, it should be understood that the terms "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.

[0085] Those skilled in the art will understand that, in addition to implementing the system, apparatus, and their modules provided by this invention in purely computer-readable program code, the same program can be implemented in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers by logically programming the method steps. Therefore, the system, apparatus, and their modules provided by this invention can be considered a hardware component, and the modules included therein for implementing various programs can also be considered structures within the hardware component; alternatively, modules for implementing various functions can be considered both software programs implementing the method and structures within the hardware component.

[0086] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.

Claims

1. A double negative acoustic metamaterial structure based on a complementary medium model, characterized in that, include: Multiple maze layers are stacked along the axial direction, with partitions between adjacent maze layers; wherein, the maze layer is a multi-layer discretized structure, and the multi-layer discretized structure forms a stepped approximation; the stepped approximation is to provide piecewise constant equivalent bulk modulus in the corresponding radial intervals of the multiple discrete maze layers, so as to approximate the continuous radial distribution of the target bulk modulus. The labyrinth layer adopts a periodic curled acoustic channel structure; the curled acoustic channel structure achieves negative equivalent mass density and negative equivalent bulk modulus by extending the sound propagation path.

2. The double negative acoustic metamaterial structure based on the complementary medium model according to claim 1, characterized in that, The maze layer includes: an outer ring boundary, multiple circumferentially distributed curled channels, a sector segmentation structure, and a central region; Wherein, the outer ring boundary is located at the outermost radial direction of the maze layer; the central region is located at the center of the maze layer; the sector segmentation structure is radially connected between the outer ring boundary and the central region, and divides the annular region between them into multiple sectors; the curled audio channel is set in each of the sectors.

3. The double negative acoustic metamaterial structure based on the complementary medium model according to claim 2, characterized in that, The central angle corresponding to the circumferential sector is: Inner diameter satisfies: in, This refers to the number of circumferential sectors. R is the outer radius of the outer ring boundary of the maze layer; i The inner radius of the central region; This refers to the width of the audio channel; Number of audio channels; The acoustic tract wall is thick.

4. The double negative acoustic metamaterial structure based on the complementary medium model according to claim 1, characterized in that, The plurality of axially stacked labyrinth layers satisfy the target bulk modulus distribution radially: in, R is the radial coordinate from the center of the maze layer; κ(r) is the target equivalent bulk modulus at that radial coordinate; o The outer radius of the outer ring boundary of the maze layer is 2R; the outer diameter of the maze layer is 2R. o .

5. A reverse design method for a double negative acoustic metamaterial structure based on a complementary medium model, characterized in that, include: Step S1: Establish a complementary medium model based on transform acoustics theory and derive the distribution of the target equivalent parameters; Step S2: Establish the mapping relationship between geometric parameters and equivalent parameters using a multilayer perceptron; Step S3: Use the beam search algorithm to generate some multi-layer candidate structures along the layer sequence from the outside to the inside; for the first layer, generate multiple single-layer candidate structures within the range of preset geometric parameters; for the i-th layer, based on the partially multi-layer candidate structures retained in the (i-1)-th layer, generate the i-th layer single-layer candidate structure under the conditions of satisfying the inner diameter, interlayer continuity and preset manufacturability constraints of the (i-1)-th layer, and combine it with its corresponding partially multi-layer candidate structures. Step S4: Use a multilayer perceptron to predict the equivalent parameters of each single-layer candidate structure, and score the expanded multilayer candidate structures based on target equivalent parameter matching, interlayer continuity constraints, terminal inner diameter constraints, and thickness uniformity constraints. The top K multi-layer candidate structures with the highest scores are retained as the candidate retention set for the current layer, and the candidate retention set is used as the constraint basis for generating the next single-layer candidate structure, until the preset number of layers or the target inner diameter is reached.

6. The reverse design method for a double negative acoustic metamaterial structure based on a complementary medium model according to claim 5, characterized in that, Step S1 includes: The target equivalent mass density satisfies: The target bulk modulus satisfies: ； in, R is the radial coordinate from the center of the maze layer; κ(r) is the target equivalent bulk modulus at that radial coordinate; o It is the outer radius of the outer ring boundary of the maze layer.

7. The reverse design method for a double negative acoustic metamaterial structure based on a complementary medium model according to claim 5, characterized in that, Step S2 includes: parametrically sampling the outer diameter, wall thickness, channel width, number of channel turns, and number of sectors; performing finite element simulation on each set of geometric parameters to obtain the corresponding equivalent mass density and equivalent bulk modulus, forming training sample pairs; training a multilayer perceptron with the geometric parameters as input and the equivalent acoustic parameters as output, and obtaining a forward surrogate model for predicting the equivalent parameters of candidate structures after training.

8. The reverse design method for a double negative acoustic metamaterial structure based on a complementary medium model according to claim 5, characterized in that, In step S4, the expanded partial multilayer candidate structures are scored based on target equivalent parameter matching, interlayer continuity constraints, terminal inner diameter constraints, and thickness uniformity constraints, including: in, A comprehensive score for a complete multi-layered structure; The local score for the candidate structure at layer i; The penalty incurred when the inner diameter of the terminal deviates from the target inner diameter; and These are the weighting coefficients for the corresponding penalty items; The penalty for uneven distribution of thickness in each layer.

9. The reverse design method for a double negative acoustic metamaterial structure based on a complementary medium model according to claim 8, characterized in that, The step of retaining the top K candidate structures based on the score and expanding the next layer of candidate structures until a preset number of layers or a target inner diameter is reached includes: If the overall score of the optimal candidate structure does not reach the preset threshold when the preset number of layers or target inner diameter is reached, the beam width K is increased and / or the sampling range of the candidate geometric parameters is adjusted before re-executing candidate generation, prediction and scoring; when the preset maximum number of iterations is reached but the threshold is still not met, the current optimal candidate structure is output and its unmet constraints are recorded for further adjustment of target parameters or creation of constraints.

10. A reverse design system for a double negative acoustic metamaterial structure based on a complementary medium model, characterized in that, include: Module M1: Establishes a complementary medium model based on transform acoustics theory and derives the target equivalent parameter distribution; Module M2: Employs a multilayer perceptron to establish a mapping relationship between geometric parameters and equivalent parameters; Module M3: A bundle search algorithm is used to generate some multi-layer candidate structures along the layer sequence from the outside to the inside. For the first layer, multiple single-layer candidate structures are generated within the range of preset geometric parameters. For the i-th layer, based on the partial multi-layer candidate structures retained in the (i-1)-th layer, a single-layer candidate structure of the i-th layer is generated under the conditions of satisfying the inner diameter, interlayer continuity and preset manufacturability constraints of the (i-1)-th layer, and combined with its corresponding partial multi-layer candidate structures. Module M4: Utilizes a multilayer perceptron to predict the equivalent parameters of each single-layer candidate structure, and scores the expanded multilayer candidate structures based on target equivalent parameter matching, interlayer continuity constraints, terminal inner diameter constraints, and thickness uniformity constraints. The top K multi-layer candidate structures with the highest scores are retained as the candidate retention set for the current layer, and the candidate retention set is used as the constraint basis for generating the next single-layer candidate structure, until the preset number of layers or the target inner diameter is reached.

Images