Acoustic metamaterial plate structure design method and system based on convolutional neural network

Abstract

The present invention relates to the field of acoustic technologies, and discloses a method and a system for designing an acoustic metamaterial plate structure based on a convolutional neural network. The method includes: obtaining a first acoustic metamaterial plate and a first acoustic metamaterial plate model respectively based on first random geometric parameters; performing finite element analysis on the first acoustic metamaterial plate model based on target acoustic performance to obtain a target finite element model; training a first convolutional neural network model respectively at different timestamps according to an augmented data set to obtain a plate structure design source model corresponding to each timestamp; calculating the similarity between the first apparent feature and each second apparent feature respectively to obtain the similarity corresponding to each plate structure design source model; and performing feature transformation on the target acoustic performance and environmental data of the acoustic metamaterial to be designed according to a target plate structure design model to obtain target geometric parameters corresponding to the acoustic metamaterial to be designed. The present invention can accurately and efficiently realize the design of an acoustic metamaterial plate structure.

Description

Technical Field

[0001] The present invention relates to the field of acoustic technologies, and particularly to a design method and system for an acoustic metamaterial plate structure based on a convolutional neural network. Background Art

[0002] As an artificially synthesized material with special acoustic properties, acoustic metamaterials can achieve special control of sound waves through their unique microstructure design, and have broad application prospects in the fields of noise control, acoustic imaging, acoustic sensors, etc. With the continuous expansion of application scenarios, the performance requirements of acoustic metamaterials are gradually increasing. How to design a metamaterial plate structure with excellent acoustic performance has become a hot research issue.

[0003] Traditional design techniques for acoustic metamaterial plate structures mostly rely on empirical formulas, physical experiments, and finite element analysis, but this technology has many disadvantages. First, the design efficiency of this method is low, and the design results are often limited to existing empirical data. Second, different acoustic metamaterials need to construct and train different models, consuming a large amount of computing resources and being difficult to accurately meet the complex and changeable actual application requirements. Finally, traditional methods are difficult to consider the influence of environmental factors on acoustic performance, resulting in certain limitations in the design results in actual applications. Summary of the Invention

[0004] In view of the above-mentioned disadvantages of the traditional technology, the present invention provides a design method and system for an acoustic metamaterial plate structure based on a convolutional neural network.

[0005] In a first aspect, an embodiment of the present invention provides a design method for an acoustic metamaterial plate structure based on a convolutional neural network, including:

[0006] Designing an acoustic metamaterial based on first random geometric parameters to obtain a first acoustic metamaterial plate and a first acoustic metamaterial plate model respectively;

[0007] Performing an acoustic performance test on the first acoustic metamaterial plate to obtain a target acoustic performance, and performing a finite element analysis on the first acoustic metamaterial plate model based on the target acoustic performance to obtain a target finite element model;

[0008] Performing data augmentation on a number of groups of second random geometric parameters based on the target finite element model to obtain an augmented data set, and training a first convolutional neural network model at different timestamps according to the augmented data set to obtain a plate structure design source model corresponding to each timestamp, where the first convolutional neural network model includes a convolutional neural network module and an attention mechanism module;

[0009] Extract the first apparent feature of the acoustic metamaterial to be designed and the second apparent feature corresponding to the acoustic metamaterial at each timestamp based on the second convolutional neural network model, and calculate the similarity between the first apparent feature and each second apparent feature respectively to obtain the similarity corresponding to each plate structure design source model, where the second convolutional neural network model includes a number of convolutional layers;

[0010] Determine the target plate structure design model corresponding to the acoustic metamaterial to be designed based on the similarity, and perform feature transformation on the target acoustic performance and environmental data of the acoustic metamaterial to be designed according to the target plate structure design model to obtain the target geometric parameters corresponding to the acoustic metamaterial to be designed.

[0011] Preferably, the obtaining of the target acoustic performance by testing the acoustic performance of the first acoustic metamaterial plate includes:

[0012] Test the acoustic performance of the first acoustic metamaterial plate to obtain the initial acoustic performance;

[0013] Obtain the environmental data of the acoustic metamaterial, and perform arithmetic processing on the environmental data based on a preset environmental function to obtain an environmental factor;

[0014] Use the environmental factor to perform weighted correction on the initial acoustic performance to obtain the target acoustic performance.

[0015] Preferably, the preset environmental function is characterized by the following formula:

[0016]

[0017] Wherein, represents the environmental factor, represents the environmental data, represents the reference environmental data, represents the environmental temperature, represents the environmental humidity, represents the dust concentration, represents the atmospheric pressure, , represents the temperature weight, represents the humidity weight, represents the dust concentration weight, represents the atmospheric pressure weight, represents the dynamic adjustment function, represents the dust concentration attenuation coefficient, represents the dust concentration adjustment factor, represents the pressure function, represents the parameter For the parameter The importance ratio of;

[0018] The dynamic adjustment function is characterized by the following formula:

[0019]

[0020] where , represents the temperature correction coefficient, represents the humidity attenuation coefficient;

[0021] The pressure function is characterized by the following formula:

[0022]

[0023] where and represent the high-pressure region correction coefficient, represents the low-pressure region attenuation coefficient.

[0024] Preferably, the finite element analysis of the first acoustic metamaterial plate model based on the target acoustic performance to obtain a target finite element model includes:

[0025] Discretize the first acoustic metamaterial plate model to obtain an initial finite element model;

[0026] Perform acoustic simulation on the initial finite element model to obtain a comparative acoustic performance;

[0027] Calculate the performance deviation between the target acoustic performance and the comparative acoustic performance;

[0028] Iteratively adjust the material properties and boundary conditions of the initial finite element model based on the performance deviation until the performance deviation reaches a convergence state to obtain a target finite element model.

[0029] Preferably, the data augmentation of several groups of second random geometric parameters based on the target finite element model to obtain an augmented data set includes:

[0030] Obtain several groups of second random geometric parameters, and design the acoustic metamaterial based on each group of the second random geometric parameters respectively to obtain corresponding second acoustic metamaterial plate models;

[0031] Obtain the environmental data when designing each of the second acoustic metamaterial plate models, and perform arithmetic processing on each group of the environmental data respectively to obtain corresponding environmental factors;

[0032] Input each of the second acoustic metamaterial plate models into the target finite element model for acoustic simulation to obtain the acoustic performance corresponding to each of the second acoustic metamaterial plate models;

[0033] Construct an augmented data set from the second random geometric parameters, the acoustic performance, and the environmental factors corresponding to each of the second acoustic metamaterial plate models.

[0034] Preferably, training the first convolutional neural network model at different timestamps based on the augmented data set to obtain a plate structure design source model corresponding to each timestamp includes:

[0035] Divide the augmented data set into a training set and a test set according to a preset ratio;

[0036] Train the first convolutional neural network model at different timestamps based on the training set to obtain a first plate structure design source model corresponding to each timestamp;

[0037] Evaluate and optimize each of the first plate structure design source models based on the test set to obtain a second plate structure design source model corresponding to each timestamp;

[0038] Characterize each of the second plate structure design source models as the plate structure design source model corresponding to the timestamp.

[0039] Preferably, based on the second convolutional neural network model, extract the first apparent feature of the acoustic metamaterial to be designed and the second apparent feature of the acoustic metamaterial corresponding to each timestamp, and calculate the similarity between the first apparent feature and each second apparent feature to obtain the similarity corresponding to each plate structure design source model, including:

[0040] Obtain the first surface image of the acoustic metamaterial to be designed and the second surface image of the acoustic metamaterial at each timestamp;

[0041] Based on the second convolutional neural network model, extract features from the first surface image and each second surface image to obtain the first apparent feature of the acoustic metamaterial to be designed and the second apparent feature of the acoustic metamaterial corresponding to each timestamp;

[0042] Calculate the cosine similarity and Pearson correlation coefficient between the first apparent feature and each second apparent feature;

[0043] Perform weighted fusion on the cosine similarity and the Pearson correlation coefficient to obtain the similarity corresponding to each plate structure design source model.

[0044] Preferably, determining the target plate structure design model corresponding to the acoustic metamaterial to be designed based on the similarity includes:

[0045] If the maximum value among the similarities is greater than a preset similarity threshold, then the board structure design source model corresponding to the maximum value is characterized as the first board structure design model corresponding to the acoustic metamaterial to be designed;

[0046] The particle swarm optimization algorithm is used to optimize the parameters of the first board structure design model to obtain a second board structure design model;

[0047] The second board structure design model is characterized as the target board structure design model corresponding to the acoustic metamaterial to be designed.

[0048] Preferably, the feature transformation of the target acoustic performance and environmental data of the acoustic metamaterial to be designed is performed according to the target board structure design model to obtain the target geometric parameters corresponding to the acoustic metamaterial to be designed, including:

[0049] An acoustic performance test is performed on the acoustic metamaterial to be designed to obtain the acoustic performance corresponding to the acoustic metamaterial to be designed;

[0050] The environmental data during the acoustic performance test of the acoustic metamaterial to be designed is obtained, and the environmental data is processed by calculation to obtain the environmental factor corresponding to the acoustic metamaterial to be designed;

[0051] Based on the environmental factor, the acoustic performance is weighted and corrected to obtain the target acoustic performance corresponding to the acoustic metamaterial to be designed;

[0052] The target board structure design model is used to perform feature extraction and mapping on the target acoustic performance and the environmental factor to obtain the target geometric parameters corresponding to the acoustic metamaterial to be designed.

[0053] In a second aspect, an embodiment of the present invention provides an acoustic metamaterial board structure design system based on a convolutional neural network, including:

[0054] A design module, configured to design an acoustic metamaterial based on first random geometric parameters to obtain a first acoustic metamaterial board and a first acoustic metamaterial board model respectively;

[0055] A finite element analysis module, configured to perform an acoustic performance test on the first acoustic metamaterial board to obtain a target acoustic performance, and perform a finite element analysis on the first acoustic metamaterial board model based on the target acoustic performance to obtain a target finite element model;

[0056] A model construction module, configured to perform data augmentation on a plurality of groups of second random geometric parameters based on the target finite element model to obtain an augmented data set, and train a first convolutional neural network model at different timestamps according to the augmented data set, so as to obtain a plate structure design source model corresponding to each timestamp, wherein the first convolutional neural network model includes a convolutional neural network module and an attention mechanism module;

[0057] A similarity calculation module, configured to respectively extract a first apparent feature of the acoustic metamaterial to be designed and a second apparent feature corresponding to each timestamp of the acoustic metamaterial based on a second convolutional neural network model, and calculate the similarity between the first apparent feature and each second apparent feature respectively, so as to obtain the similarity corresponding to each plate structure design source model, wherein the second convolutional neural network model includes a plurality of convolutional layers;

[0058] A geometric parameter determination module, configured to determine a target plate structure design model corresponding to the acoustic metamaterial to be designed based on the similarity, and perform feature transformation on the target acoustic performance and environmental data of the acoustic metamaterial to be designed according to the target plate structure design model, so as to obtain the target geometric parameters corresponding to the acoustic metamaterial to be designed.

[0059] Compared with the prior art, an acoustic metamaterial plate structure design method and system based on a convolutional neural network according to an embodiment of the present invention have the beneficial effects that: by fusing multi-scale environmental data features to correct the acoustic performance, a reliable basis is provided for subsequent design; a data augmentation technology is adopted to generate a large amount of reliable training data, and multiple plate structure design source models are trained based on the augmented data set, effectively improving the generalization ability and adaptability of the model; by calculating the similarity to determine the target plate structure design model, the most suitable model can be quickly and accurately matched for the acoustic metamaterial to be designed, improving the design efficiency and accuracy; based on the convolutional neural network, the geometric parameters of the acoustic metamaterial are accurately designed, realizing the accurate mapping from the performance and environmental data to the structural geometric parameters, and providing strong technical support for the research and development and application of the acoustic metamaterial. Description of the Drawings

[0060] Figure 1 is a schematic flowchart of an acoustic metamaterial plate structure design method based on a convolutional neural network according to an embodiment of the present invention;

[0061] Figure 2 is a schematic flowchart of a process for obtaining the target acoustic performance and the target finite element model according to an embodiment of the present invention;

[0062] Figure 3 is a schematic flowchart of a process for calculating the similarity according to an embodiment of the present invention;

[0063] Figure 4It is a schematic structural diagram of a system for designing the structure of an acoustic metamaterial plate based on a convolutional neural network according to an embodiment of the present invention. Specific embodiments

[0064] The following will further describe in detail the specific embodiments of the present invention in conjunction with the accompanying drawings and embodiments. The following embodiments are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0065] In the description of the present invention, it should be understood that the terms "first" and "second" etc. used in the present invention are used to distinguish different objects, rather than to describe a specific order.

[0066] In the description of the present invention, it should be noted that unless otherwise defined, all technical and scientific terms used in the present invention have the same meaning as commonly understood by those skilled in the art. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments, and are not intended to limit the present invention. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.

[0067] As Figure 1 shown, an embodiment of the present invention provides a method for designing the structure of an acoustic metamaterial plate based on a convolutional neural network, including the steps:

[0068] S1. Design an acoustic metamaterial based on the first random geometric parameters to obtain a first acoustic metamaterial plate and a first acoustic metamaterial plate model respectively;

[0069] Take a set of random geometric parameters as the first random geometric parameters, and manufacture the first acoustic metamaterial plate according to the first random geometric parameters and the acoustic metamaterial. Among them, the geometric parameters include plate thickness, pore size, pore spacing, and arrangement pattern. Specifically, the first random geometric parameters in this embodiment are a plate thickness of 15 mm, a pore diameter of 4 mm, a pore spacing of 20 mm, and a square arrangement.

[0070] In the modeling software, create a three-dimensional geometric model of the first acoustic metamaterial plate, that is, the first acoustic metamaterial plate model, according to the first random geometric parameters. Among them, input the material properties of the acoustic metamaterial into the model so that the model can accurately simulate the response of the acoustic metamaterial plate under the action of sound waves. The setting of boundary conditions will affect the propagation and emission of sound waves in the model, which is crucial for accurately simulating acoustic performance. Specifically, the material properties include material density, elastic modulus, damping coefficient, material sound velocity, and thermal expansion coefficient, and the boundary conditions include fixed boundary, free boundary, and periodic boundary. Further, the material properties in this embodiment are a material density of 2.6 g / cm 3 , an elastic modulus of 100 GPa, a damping coefficient of 0.05, a material sound velocity of 3000 m / s, and a thermal expansion coefficient of 1*10 -5 / °C.

[0071] S2. Perform acoustic performance tests on the first acoustic metamaterial plate to obtain the target acoustic performance, and perform finite element analysis on the first acoustic metamaterial plate model based on the target acoustic performance to obtain the target finite element model;

[0072] Specifically, as Figure 2 shown, step S2 includes:

[0073] S201. Perform acoustic performance tests on the first acoustic metamaterial plate to obtain the initial acoustic performance;

[0074] Perform acoustic performance tests on the first acoustic metamaterial plate to obtain the initial acoustic performance at a specific frequency. Specifically, the initial acoustic performance includes the sound absorption coefficient, acoustic impedance, and sound transmission loss. In this embodiment, acoustic performance tests are performed on the first acoustic metamaterial plate made of the first random geometric parameters to obtain the initial acoustic performance at 1000 Hz, and the initial acoustic performance is a sound absorption coefficient of 0.45, an acoustic impedance of 1500 + 200ikg / (m 2 *s), and a sound transmission loss of 15 dB.

[0075] S202. Obtain the environmental data of the acoustic metamaterial, and perform arithmetic processing on the environmental data based on a preset environmental function to obtain the environmental factor;

[0076] Collect environmental data through an environmental sensing network, and input the environmental data into a preset environmental function to obtain the environmental factor.

[0077] Specifically, the preset environmental function is characterized by the following formula:

[0078]

[0079] where represents the environmental factor, represents the environmental data, represents the reference environmental data, represents the environmental temperature, represents the environmental humidity, represents the dust concentration, represents the atmospheric pressure, , represents the temperature weight, represents the humidity weight, represents the dust concentration weight, represents the atmospheric pressure weight, represents the dynamic adjustment function, represents the dust concentration attenuation coefficient, represents the dust concentration adjustment factor, represents the pressure function, represents the parameter The importance ratio of the parameter is as follows

[0080] Furthermore, the dynamic adjustment function is characterized by the following formula:

[0081]

[0082] where , represents the temperature correction coefficient, represents the humidity attenuation coefficient

[0083] Furthermore, the pressure function is characterized by the following formula:

[0084]

[0085] where and represent the correction coefficient of the high-pressure area, represents the attenuation coefficient of the low-pressure area

[0086] In this embodiment, the environmental temperature collected by the environmental sensing network is 28 °C (reference temperature 25 °C), the environmental humidity is 55% (reference humidity 50%), the dust concentration is 60 μg / m 3 (reference dust concentration 50 μg / m 3 ), and the atmospheric pressure is 100.5 kPa (reference atmospheric pressure 100 kPa); the temperature weight is 0.3, the humidity weight is 0.2, the dust concentration weight is 0.7, the atmospheric pressure weight is 0.4, the temperature correction coefficient is 0.1, the humidity attenuation coefficient is 0.05, the dust concentration attenuation coefficient is 0.9, the dust concentration adjustment factor is 60 μg 2 / m 6 , and the importance ratios , , , , , are 4, 5, 3, 3.5, 4.5, 2.5 respectively. Based on the above calculations, the environmental factor is 1.12598

[0087] S203. The initial acoustic performance is weighted and corrected by the environmental factor to obtain the target acoustic performance

[0088] That is to say, the target acoustic performance is the product of the initial acoustic performance and the environmental factor

[0089] Specifically, the target acoustic performance in this embodiment is an absorption coefficient of 0.507, an acoustic impedance of 1688.97 + 225.2i kg / (m 2 *s), and a sound transmission loss of 16.89 dB

[0090] Further, as Figure 2 shown, step S2 further includes:

[0091] S204. Discretize the first acoustic metamaterial plate model to obtain an initial finite element model;

[0092] Determine the material properties and boundary conditions of the first acoustic metamaterial plate model, and input the first acoustic metamaterial plate model, material properties, and boundary conditions into finite element software for discretization to obtain an initial finite element model. Specifically, the finite element software divides the originally continuous three-dimensional geometric model into a finite number of elements according to certain rules, determines the connection points between the elements, and assigns the material properties and boundary conditions to these elements and nodes, thereby transforming the complex physical model into a finite element model that can perform numerical calculations, and realizing the simulation analysis of the acoustic metamaterial plate.

[0093] S205. Perform acoustic simulation on the initial finite element model to obtain comparative acoustic performance;

[0094] Perform acoustic simulation on the initial finite element model using a standard sound wave to obtain comparative acoustic performance.

[0095] S206. Calculate the performance deviation between the target acoustic performance and the comparative acoustic performance;

[0096] Since the acoustic performance includes three evaluation indicators: sound absorption coefficient, acoustic impedance, and sound transmission loss, the performance deviation is represented by the weighted result of the deviation of each indicator. Specifically, in this embodiment, the following formula is used to calculate the performance deviation:

[0097]

[0098] where, represents the performance deviation, represents the sound absorption coefficient deviation, represents the acoustic impedance deviation, represents the sound transmission loss deviation. It should be noted that the weight coefficient of the deviation of each indicator can be adjusted according to the actual situation, and is not limited to 2, 0.0001, and 0.1 set in this embodiment.

[0099] S207. Iteratively adjust the material properties and boundary conditions of the initial finite element model based on the performance deviation until the performance deviation reaches a convergence state to obtain a target finite element model.

[0100] Iteratively adjust the material properties and boundary conditions of the initial finite element model based on the performance deviation until the performance deviation is less than the preset deviation threshold or no longer decreases, and stop the iterative adjustment to obtain a target finite element model. The preset deviation threshold in this embodiment is 0.3.

[0101] S3. Augment the data for several groups of second random geometric parameters based on the target finite element model to obtain an augmented dataset, and train the first convolutional neural network model based on the augmented dataset at different timestamps respectively to obtain the source models for the plate structure design corresponding to each timestamp;

[0102] Specifically, step S3 includes:

[0103] 1) Obtain several groups of second random geometric parameters, and design acoustic metamaterials based on each group of second random geometric parameters respectively to obtain the corresponding second acoustic metamaterial plate models;

[0104] Take several groups of random geometric parameters as several groups of second random geometric parameters. In the modeling software, create a three-dimensional geometric model of the corresponding second acoustic metamaterial plate, that is, the second acoustic metamaterial plate model, according to each group of second random geometric parameters. Specifically, in this embodiment, 10 groups of second random geometric parameters are obtained, that is, 10 second acoustic metamaterial plate models are constructed.

[0105] 2) Obtain the environmental data when designing each second acoustic metamaterial plate model, and perform arithmetic processing on each group of environmental data respectively to obtain the corresponding environmental factors;

[0106] Collect the environmental data when designing each second acoustic metamaterial plate model through the environmental sensing network, and input each group of environmental data into the preset environmental function respectively to obtain the corresponding environmental factors. It should be noted that the process of calculating the environmental factors through the preset environmental function can be referred to the previous text and will not be elaborated here one by one.

[0107] 3) Input each second acoustic metamaterial plate model into the target finite element model for acoustic simulation respectively to obtain the acoustic performance corresponding to each second acoustic metamaterial plate model;

[0108] 4) Combine the second random geometric parameters, acoustic performance, and environmental factors corresponding to each second acoustic metamaterial plate model to form an augmented dataset;

[0109] The augmented dataset obtained by the data augmentation technology in this embodiment includes 10 groups of augmented data, and each group of augmented data includes the corresponding second random geometric parameters, acoustic performance, and environmental factors.

[0110] Furthermore, step S3 also includes:

[0111] 5) Divide the augmented dataset into a training set and a test set according to a preset ratio;

[0112] In this embodiment, the random forest algorithm is used to divide the augmented dataset into a training set and a test set according to 8:2.

[0113] 6) Train the first convolutional neural network model based on the training set at different timestamps respectively to obtain the first plate structure design source model corresponding to each timestamp;

[0114] The following is a specific description of the first convolutional neural network model:

[0115] The first convolutional neural network model includes an input layer, a fully connected network module, a convolutional neural network module, an attention mechanism module, and an output layer. The input layer is used to receive input data such as geometric parameters, acoustic performance, and environmental factors; the fully connected network module is used to map acoustic performance and environmental factors to a high-dimensional space and extract high-level features; the convolutional neural network module is used to reshape the high-level feature vector into a two-dimensional feature map and extract spatial features, capture the local dependence between geometric parameters, and output a convolutional feature map; the attention mechanism module is used to enhance the key features in the convolutional feature map and suppress irrelevant features, capture the global dependence between geometric parameters, and output a weighted feature vector; the output layer is used to map the weighted feature vector to the geometric parameter space and output the designed geometric parameters.

[0116] Use the training set to train the first convolutional neural network model at different timestamps respectively, and represent the trained first convolutional neural network model as the first plate structure design source model corresponding to each timestamp.

[0117] 7) Based on the test set, evaluate and optimize each first plate structure design source model respectively to obtain the second plate structure design source model corresponding to each timestamp;

[0118] Use the test set to evaluate the first plate structure design source model, optimize the model according to the evaluation results, continuously adjust the parameters of the model, gradually improve the performance of the model, and finally obtain the target model, that is, the second plate structure design source model, so that it can output the designed geometric parameters more accurately.

[0119] 8) Represent each second plate structure design source model as the plate structure design source model corresponding to the corresponding timestamp.

[0120] In this embodiment, each plate structure design source model is stored in the source model library according to the timestamp.

[0121] S4. Based on the second convolutional neural network model, extract the first apparent feature of the acoustic metamaterial to be designed and the second apparent feature of the acoustic metamaterial corresponding to each timestamp respectively, and calculate the similarity between the first apparent feature and each second apparent feature respectively to obtain the similarity corresponding to each plate structure design source model;

[0122] Specifically, as Figure 3 shown, step S4 includes:

[0123] S401. Obtain the first surface image of the acoustic metamaterial to be designed and the second surface images of the acoustic metamaterial at each time stamp;

[0124] Obtain the first surface image of the acoustic metamaterial to be designed and the second surface images of the acoustic metamaterial at each time stamp through a high-definition camera. Meanwhile, perform spectral analysis on the acoustic metamaterial to be designed and the acoustic metamaterial at each time stamp respectively to obtain the first spectral feature of the acoustic metamaterial to be designed and the second spectral features of the acoustic metamaterial at each time stamp. In addition, obtain the first hardness and the first volume density of the acoustic metamaterial to be designed, and the second hardness and the second volume density of the acoustic metamaterial at each time stamp. It can be understood that the acoustic metamaterial to be designed and the acoustic metamaterial belong to different materials of the same type.

[0125] S402. Based on the second convolutional neural network model, perform feature extraction on the first surface image and each second surface image respectively to obtain the first apparent feature of the acoustic metamaterial to be designed and the second apparent features corresponding to the acoustic metamaterial at each time stamp;

[0126] The following is a specific description of the second convolutional neural network model:

[0127] Use the second convolutional neural network model to process the surface image to obtain the image surface features. Specifically, the second convolutional neural network model is composed of an input layer, a first convolutional layer, a first pooling layer, a second convolutional layer, a second pooling layer, a third convolutional layer, a third pooling layer, a fourth convolutional layer, a fourth pooling layer, a first fully connected layer, a second fully connected layer, and an output layer connected in series in sequence. The input layer preprocesses the surface image to obtain a surface-enhanced image; the first convolutional layer extracts the texture features of the surface-enhanced image using the gray-level co-occurrence matrix and local binary pattern; the second convolutional layer extracts the roughness and surface defect features of the surface-enhanced image using adaptive median filtering and edge detection; the third convolutional layer further extracts the high-level texture and structural features of the surface-enhanced image; the fourth convolutional layer extracts the thickness uniformity features of the surface-enhanced image using a three-dimensional reconstruction algorithm; the first fully connected layer extracts RGB features using the color histogram and color moment; the second fully connected layer calculates and extracts the specular reflectance features using the specular highlight detection algorithm and reflectance algorithm; the output layer receives the image processing results of the fully connected layer and splices the image features to output the image surface features.

[0128] Splice the image surface features, the first spectral feature, the first hardness, and the first volume density of the acoustic metamaterial to be designed to obtain the first apparent feature of the acoustic metamaterial to be designed, and splice the image surface features, the second spectral feature, the second hardness, and the second volume density of the acoustic metamaterial to obtain the second apparent features corresponding to the acoustic metamaterial at each time stamp.

[0129] S403. Calculate the cosine similarity and Pearson correlation coefficient between the first apparent feature and each second apparent feature;

[0130] S404. Perform weighted fusion on the cosine similarity and Pearson correlation coefficient to obtain the similarity corresponding to each plate structure design source model.

[0131] It can be understood that the similarity is actually a comprehensive similarity, which is obtained by weighted fusion of the cosine similarity and Pearson correlation coefficient of the first apparent feature and the second apparent feature.

[0132] S5. Based on the similarity, determine the target plate structure design model corresponding to the acoustic metamaterial to be designed, and perform feature transformation on the target acoustic performance and environmental data of the acoustic metamaterial to be designed according to the target plate structure design model to obtain the target geometric parameters corresponding to the acoustic metamaterial to be designed.

[0133] Specifically, step S5 includes:

[0134] 1) If the maximum value in the similarity is greater than the preset similarity threshold, then characterize the plate structure design source model corresponding to the maximum value as the first plate structure design model corresponding to the acoustic metamaterial to be designed;

[0135] When the highest comprehensive similarity is greater than the preset similarity threshold, take the corresponding plate structure design source model in the source model library as the first plate structure design model. Among them, the preset similarity threshold is set to 0.7 in this embodiment. On the contrary, if it is determined that there is no plate structure design source model for the acoustic metamaterial to be designed that can be transferred for learning, then the above data augmentation steps need to be continued to increase the training data to enrich the source model library.

[0136] 2) Use the particle swarm optimization algorithm to optimize the parameters of the first plate structure design model to obtain the second plate structure design model;

[0137] Before optimizing the parameters of the first plate structure design model, it is necessary to determine the optimization objective function. The following is an explanation of the construction process of the optimization objective function:

[0138] i) Obtain the geometric parameter deviation

[0139] Collect the environmental data of the acoustic metamaterial to be designed through the environmental sensing network, and input the environmental data into the preset environmental function to obtain the environmental factor. Another set of random geometric parameters is taken as the third random geometric parameter, and the acoustic metamaterial to be designed is made into an acoustic metamaterial plate according to the third random geometric parameter, and the acoustic performance of the acoustic metamaterial plate is tested to obtain the acoustic performance. Input the acoustic performance and the environmental factor into the first plate structure design model to obtain the geometric parameters, and calculate the geometric parameter deviation between the third random geometric parameter and the geometric parameters.

[0140] ii) Determine the optimization objective function

[0141] Determine the optimization objective function according to the geometric parameter deviation. Specifically, the following formula is used to characterize the optimization objective function:

[0142]

[0143]

[0144] where, represents the optimization objective function, represents the deviation function, represents the geometric dimension deviation, represents the dimension weight, and represent the dimension penalty coefficient, represents the geometric parameter, represents the third random geometric parameter, represents the penalty weight, represents the quality weight, represents the similarity weight, represents the structural equivalent stress related to the geometric dimension deviation, represents the allowable structural stress, represents the weight of the acoustic metamaterial per unit area corresponding to the geometric parameter, represents the weight of the acoustic metamaterial per unit area corresponding to the third random geometric parameter, represents the volume density of the acoustic metamaterial to be designed, represents the unit area, represents the plate thickness of the geometric parameter, represents the plate thickness of the third random geometric parameter, represents the Pearson correlation coefficient between the third random geometric parameter vector and the geometric parameter vector.

[0145] Use the particle swarm optimization algorithm to optimize the hyperparameters of the first plate structure design model, generate two chaotic sequences at two different scales, and perform chaotic mapping on the population as follows:

[0146]

[0147] where, represents the chaotic value of the th dimension, represents the chaotic value of the th dimension corresponding to the first chaotic sequence, represents the chaotic value of the th dimension corresponding to the second chaotic sequence, and represent the scale weight, represents the The chaotic value of the dimension, represents a chaotic random number, represents a control parameter used to adjust the chaotic sequence distribution.

[0148] Calculate the fitness of the particle and determine the optimal position of the particle at the th iteration in the dimension and the optimal position of the population , and update the particle position and velocity as follows:

[0149]

[0150]

[0151] where, represents the updated position of the particle at the th iteration in the dimension , represents the position of the particle at the th iteration in the dimension , represents the updated velocity of the particle at the th iteration in the dimension , represents the velocity of the particle at the th iteration in the dimension , represents the dynamically adjusted weight at the th iteration, represents the golden sine coefficient, represents the maximum number of iterations, represents the Levy flight coefficient, represents the individual learning factor, represents the social learning factor, represents a random number, and represents a random number, represents the Levy flight strategy, represents the adaptive flight step size, represents the regularization coefficient, represents the current population vector, represents the mean of the current population vector, represents the Euclidean norm.

[0152] Dynamically adjust the weight Introduce the population diversity index and convergence rate, which are calculated by the fuzzy inference system as follows:

[0153]

[0154]

[0155]

[0156] Among them, 、 and represent constant coefficients, represents the membership function of the population diversity index of represents the convergence rate of represents the number of particles, represents the population at in the dimension the average position at the iteration, represents the change in the global optimal solution in the last 5 generations of the convergence rate,

[0157] Adopt the elite reverse mutation strategy to mutate the population that has not changed globally for 5 consecutive iterations, and adopt the quantum tunneling mechanism to update the particles trapped in the local optimum, as follows:

[0158]

[0159]

[0160] Among them, represents the position of the particle mutation in the dimension after mutation, represents the particle trapped in the local optimum at in the dimension the position after the iteration update, and represent the lower and upper bounds of the variable in the dimension search space, represents the reverse learning coefficient, represents the Cauchy random distribution function, represents the standard deviation of the particle position, and represent uniform random numbers.

[0161] Calculate the fitness of the particles to determine the optimal position of the particles and the optimal position of the population, and update the particle position and velocity. Repeat the iteration until the optimization objective function is minimized and the iteration stops, and determine the optimal model parameters to output the second plate structure design model.

[0162] The size weight in the optimization objective function of this embodiment Take , the penalty weight Take 0.1, the quality weight Take 0.2, the similarity weight Take 0.1, the size penalty coefficient and Take 2 and 1.5; Set the number of particles N in the particle swarm to 50, the maximum number of iterations K to 100, the scale weight and Take 0.6 and 0.4, the chaotic random number Take 0.5, the control parameter Take 0.8; And update the particle position and velocity, repeat the iteration (when k = 20, is 1.2, when k = 40, is 0.8, when k = 60, is 0.65, when k = 65, is 0.64, when k = 6, is 0.65) until the optimization objective function reaches the minimum and no longer changes, then stop the iteration. Take the model parameters corresponding to the 65th iteration as the optimal model parameters and output the second plate structure design model.

[0163] 3) Characterize the second plate structure design model as the target plate structure design model corresponding to the acoustic metamaterial to be designed;

[0164] Furthermore, step S5 further includes:

[0165] 4) Conduct an acoustic performance test on the acoustic metamaterial to be designed to obtain the acoustic performance corresponding to the acoustic metamaterial to be designed;

[0166] 5) Obtain the environmental data during the acoustic performance test of the acoustic metamaterial to be designed, and perform arithmetic processing on the environmental data to obtain the environmental factor corresponding to the acoustic metamaterial to be designed;

[0167] 6) Based on the environmental factor, perform weighted correction on the acoustic performance to obtain the target acoustic performance corresponding to the acoustic metamaterial to be designed;

[0168] It should be noted that the execution process of steps 4) to 6) can refer to steps S201 to S203 described above, and will not be elaborated here one by one.

[0169] 7) Use the target plate structure design model to perform feature extraction and mapping on the target acoustic performance and the environmental factor to obtain the target geometric parameters corresponding to the acoustic metamaterial to be designed.

[0170] It can be understood that both the convolutional neural network module and the attention mechanism module in the target board structure design model belong to feature extraction. Specifically, in this embodiment, the target acoustic performance (absorption coefficient of 0.55, acoustic impedance of 1700 + 250ikg / m 2 *s, and sound transmission loss of 10 dB) and the environmental factor of 1.089 are input into the target board structure design model for feature extraction and mapping to obtain the target geometric parameters (plate thickness of 5 mm, hole diameter of 2 mm, hole pitch of 10 mm, hexagonal arrangement).

[0171] In an embodiment of the present invention, a method for designing an acoustic metamaterial board structure based on a convolutional neural network corrects the acoustic performance by fusing multi-scale environmental data features, providing a reliable basis for subsequent designs; uses data augmentation technology to generate a large number of reliable training data, and trains multiple board structure design source models based on the augmented data set, effectively improving the generalization ability and adaptability of the models; determines the target board structure design model through similarity calculation, which can quickly and accurately match the most suitable model for the acoustic metamaterial to be designed, improving the design efficiency and accuracy; precisely designs the geometric parameters of the acoustic metamaterial based on the convolutional neural network, realizing the accurate mapping from performance and environmental data to structural geometric parameters, and providing strong technical support for the research and development and application of acoustic metamaterials.

[0172] Based on the above method for designing an acoustic metamaterial board structure based on a convolutional neural network, as Figure 4 shown, an embodiment of the present invention provides an acoustic metamaterial board structure design system based on a convolutional neural network, including:

[0173] A design module 1, configured to design an acoustic metamaterial based on the first random geometric parameters to obtain a first acoustic metamaterial board and a first acoustic metamaterial board model respectively;

[0174] A finite element analysis module 2, configured to perform acoustic performance tests on the first acoustic metamaterial board to obtain the target acoustic performance, and perform finite element analysis on the first acoustic metamaterial board model based on the target acoustic performance to obtain a target finite element model;

[0175] A model construction module 3, configured to perform data augmentation on a number of groups of second random geometric parameters based on the target finite element model to obtain an augmented data set, and train the first convolutional neural network model at different timestamps according to the augmented data set to obtain a board structure design source model corresponding to each timestamp, where the first convolutional neural network model includes a convolutional neural network module and an attention mechanism module;

[0176] A similarity calculation module 4 is configured to respectively extract a first apparent feature of the acoustic metamaterial to be designed and a second apparent feature corresponding to the acoustic metamaterial at each time stamp based on a second convolutional neural network model, and calculate the similarity between the first apparent feature and each second apparent feature respectively to obtain the similarity corresponding to each plate structure design source model. The second convolutional neural network model includes a plurality of convolutional layers;

[0177] A geometric parameter determination module 5 is configured to determine a target plate structure design model corresponding to the acoustic metamaterial to be designed based on the similarity, and perform feature transformation on the target acoustic performance and environmental data of the acoustic metamaterial to be designed according to the target plate structure design model to obtain the target geometric parameters corresponding to the acoustic metamaterial to be designed.

[0178] It should be noted that each module in the above acoustic metamaterial plate structure design system based on a convolutional neural network can be implemented in whole or in part by software, hardware, and their combination. The above modules can be embedded in the processor of the computer device in hardware form or be independent of it, or can be stored in the memory of the computer device in software form, so as to facilitate the processor to call and execute the operations corresponding to the above modules. For the specific limitations of an acoustic metamaterial plate structure design system based on a convolutional neural network, refer to the limitations on an acoustic metamaterial plate structure design method based on a convolutional neural network in the above text. The two have the same functions and effects, and will not be elaborated here.

[0179] In summary, in the embodiment of the present invention, an acoustic metamaterial plate structure design method and system based on a convolutional neural network correct the acoustic performance by fusing multi-scale environmental data features, providing a reliable basis for subsequent designs; adopt a data augmentation technology to generate a large number of reliable training data, and train multiple plate structure design source models based on the augmented data set, effectively improving the generalization ability and adaptability of the models; determine the target plate structure design model through similarity calculation, which can quickly and accurately match the most suitable model for the acoustic metamaterial to be designed, improving the design efficiency and accuracy; accurately design the geometric parameters of the acoustic metamaterial based on the convolutional neural network, realizing the accurate mapping from performance and environmental data to structural geometric parameters, and providing strong technical support for the research and development and application of acoustic metamaterials.

[0180] Each embodiment in this specification is described in a progressive manner. For the parts that are the same or similar in each embodiment, reference can be made to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and for the relevant parts, reference can be made to the partial description of the method embodiment. It should be noted that the technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this specification.

[0181] The above description is only the preferred embodiment of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the technical principle of the present invention, several improvements and replacements can still be made, and these improvements and replacements should also be regarded as the protection scope of the present invention.

Claims

1. A method for designing an acoustic metamaterial plate structure based on a convolutional neural network, characterized in that Including: Designing an acoustic metamaterial based on a first random geometric parameter to obtain a first acoustic metamaterial plate and a first acoustic metamaterial plate model respectively; Performing an acoustic performance test on the first acoustic metamaterial plate to obtain a target acoustic performance, and performing finite element analysis on the first acoustic metamaterial plate model based on the target acoustic performance to obtain a target finite element model; Performing data augmentation on a number of groups of second random geometric parameters based on the target finite element model, and training a first convolutional neural network model at different timestamps according to the augmented dataset to obtain a plate structure design source model corresponding to each timestamp, wherein the first convolutional neural network model includes a convolutional neural network module and an attention mechanism module; Extracting a first apparent feature of the acoustic metamaterial to be designed and a second apparent feature corresponding to each timestamp of the acoustic metamaterial based on a second convolutional neural network model, and calculating the similarity between the first apparent feature and each second apparent feature respectively to obtain a similarity corresponding to each plate structure design source model, wherein the second convolutional neural network model includes a number of convolutional layers; Determining a target plate structure design model corresponding to the acoustic metamaterial to be designed based on the similarity, and performing feature transformation on the target acoustic performance and environmental data of the acoustic metamaterial to be designed according to the target plate structure design model to obtain a target geometric parameter corresponding to the acoustic metamaterial to be designed.

2. The acoustic metamaterial plate structure design method according to claim 1, wherein The performing an acoustic performance test on the first acoustic metamaterial plate to obtain a target acoustic performance includes: Performing an acoustic performance test on the first acoustic metamaterial plate to obtain an initial acoustic performance; Obtaining environmental data of the acoustic metamaterial, and performing arithmetic processing on the environmental data based on a preset environmental function to obtain an environmental factor; Using the environmental factor to perform weighted correction on the initial acoustic performance to obtain a target acoustic performance.

3. The acoustic metamaterial plate structure design method according to claim 2, wherein Characterizing the preset environmental function with the following formula: Among them, represents an environmental factor, represents environmental data, represents reference environmental data, represents the environmental temperature, represents the environmental humidity, represents the dust concentration, represents the atmospheric pressure, , represents the temperature weight, represents the humidity weight, represents the dust concentration weight, represents the atmospheric pressure weight, represents the dynamic adjustment function, represents the dust concentration attenuation coefficient, represents the dust concentration adjustment factor, represents the pressure function, represents a parameter the importance ratio of parameter to parameter represents a parameter the importance ratio of parameter to parameter Characterizing the dynamic adjustment function with the following formula: Among them, , represents the temperature correction coefficient, represents the humidity attenuation coefficient; Characterizing the pressure function with the following formula: Among them, and represent the correction coefficient for the high-pressure area, represents the attenuation coefficient for the low-pressure area.

4. The acoustic metamaterial plate structure design method according to claim 1, characterized in that The performing finite element analysis on the first acoustic metamaterial plate model based on the target acoustic performance to obtain a target finite element model includes: Performing discretization processing on the first acoustic metamaterial plate model to obtain an initial finite element model; Performing acoustic simulation on the initial finite element model to obtain a comparative acoustic performance; Calculating the performance deviation between the target acoustic performance and the comparative acoustic performance; Iteratively adjusting the material properties and boundary conditions of the initial finite element model based on the performance deviation until the performance deviation reaches a convergence state to obtain a target finite element model.

5. The method for designing an acoustic metamaterial plate structure according to claim 1, wherein The performing data augmentation on a number of groups of second random geometric parameters based on the target finite element model to obtain an augmented dataset includes: Obtaining a number of groups of second random geometric parameters, and designing the acoustic metamaterial based on each group of the second random geometric parameters respectively to obtain corresponding second acoustic metamaterial plate models; Obtain the environmental data when designing each of the second acoustic metamaterial plate models, and perform arithmetic processing on each set of the environmental data respectively to obtain the corresponding environmental factors; Input each of the second acoustic metamaterial plate models into the target finite element model for acoustic simulation respectively to obtain the acoustic performance corresponding to each of the second acoustic metamaterial plate models; Construct an augmented data set with the second random geometric parameters, the acoustic performance, and the environmental factors corresponding to each of the second acoustic metamaterial plate models.

6. The acoustic metamaterial plate structure design method according to claim 1, characterized in that Training the first convolutional neural network model at different timestamps according to the augmented data set to obtain the plate structure design source model corresponding to each timestamp, including: Divide the augmented data set into a training set and a test set according to a preset ratio; Train the first convolutional neural network model at different timestamps based on the training set to obtain the first plate structure design source model corresponding to each timestamp; Evaluate and optimize each of the first plate structure design source models based on the test set to obtain the second plate structure design source model corresponding to each timestamp; Characterize each of the second plate structure design source models as the plate structure design source model corresponding to the corresponding timestamp.

7. The acoustic metamaterial plate structure design method according to claim 1, characterized in that Based on the second convolutional neural network model, extract the first apparent feature of the acoustic metamaterial to be designed and the second apparent feature corresponding to the acoustic metamaterial at each timestamp, and calculate the similarity between the first apparent feature and each of the second apparent features respectively to obtain the similarity corresponding to each plate structure design source model, including: Obtain the first surface image of the acoustic metamaterial to be designed and the second surface image of the acoustic metamaterial at each timestamp; Based on the second convolutional neural network model, perform feature extraction on the first surface image and each of the second surface images respectively to obtain the first apparent feature of the acoustic metamaterial to be designed and the second apparent feature corresponding to the acoustic metamaterial at each timestamp; Calculate the cosine similarity and Pearson correlation coefficient between the first apparent feature and each of the second apparent features; Perform weighted fusion on the cosine similarity and the Pearson correlation coefficient to obtain the similarity corresponding to each plate structure design source model.

8. The acoustic metamaterial plate structure design method according to claim 1, wherein Determine the target plate structure design model corresponding to the acoustic metamaterial to be designed based on the similarity, including: If the maximum value in the similarity is greater than the preset similarity threshold, characterize the plate structure design source model corresponding to the maximum value as the first plate structure design model corresponding to the acoustic metamaterial to be designed; Use the particle swarm optimization algorithm to optimize the parameters of the first plate structure design model to obtain the second plate structure design model; Characterize the second plate structure design model as the target plate structure design model corresponding to the acoustic metamaterial to be designed.

9. The acoustic metamaterial plate structure design method according to claim 1, characterized in that According to the target plate structure design model, perform feature transformation on the target acoustic performance and environmental data of the acoustic metamaterial to be designed to obtain the target geometric parameters corresponding to the acoustic metamaterial to be designed, including: Perform acoustic performance tests on the acoustic metamaterial to be designed to obtain the acoustic performance corresponding to the acoustic metamaterial to be designed; Obtain the environmental data during the acoustic performance tests on the acoustic metamaterial to be designed, and perform arithmetic processing on the environmental data to obtain the environmental factor corresponding to the acoustic metamaterial to be designed; Based on the environmental factor, perform weighted correction on the acoustic performance to obtain the target acoustic performance corresponding to the acoustic metamaterial to be designed; Use the target plate structure design model to perform feature extraction and mapping on the target acoustic performance and the environmental factor to obtain the target geometric parameters corresponding to the acoustic metamaterial to be designed.

10. An acoustic metamaterial plate structure design system based on a convolutional neural network, characterized in that, Including: A design module for designing an acoustic metamaterial based on the first random geometric parameters to obtain a first acoustic metamaterial plate and a first acoustic metamaterial plate model respectively; A finite element analysis module for performing acoustic performance tests on the first acoustic metamaterial plate to obtain the target acoustic performance, and performing finite element analysis on the first acoustic metamaterial plate model based on the target acoustic performance to obtain the target finite element model; A model construction module for performing data augmentation on a number of groups of second random geometric parameters based on the target finite element model to obtain an augmented data set, and training the first convolutional neural network model at different timestamps according to the augmented data set to obtain the plate structure design source model corresponding to each timestamp, where the first convolutional neural network model includes a convolutional neural network module and an attention mechanism module; A similarity calculation module for respectively extracting the first apparent feature of the acoustic metamaterial to be designed and the second apparent feature of the acoustic metamaterial corresponding to each timestamp based on the second convolutional neural network model, and respectively calculating the similarity between the first apparent feature and each second apparent feature to obtain the similarity corresponding to each plate structure design source model, where the second convolutional neural network model includes a number of convolutional layers; A geometric parameter determination module for determining the target plate structure design model corresponding to the acoustic metamaterial to be designed based on the similarity, and performing feature conversion on the target acoustic performance and environmental data of the acoustic metamaterial to be designed according to the target plate structure design model to obtain the target geometric parameters corresponding to the acoustic metamaterial to be designed.

Images