Production method of light-transmitting sound-absorbing module suitable for gas film structure

Abstract

The invention provides a production method of a light-transmitting sound-absorbing module suitable for an air film structure, and belongs to the technical field of building sound absorption and noise reduction. Then borate flame retardant treatment is carried out; forming a plate-shaped sound absorber by adopting a hot press molding process; then micro-pore distribution parameters are calculated, and the transparent micro-pore polymer film is compounded with the sound absorber through ultrasonic welding; performing waterproof and mildew-proof treatment on a fluorocarbon water repellent and a quaternary ammonium salt mildew preventive; after accurate drying, the acoustic performance of the composite material is analyzed by applying a pre-trained neural network model, a cutting scheme is optimized, and hot edge sealing treatment is carried out; and finally, testing the sound absorption performance of the product at the frequency band of 100Hz to 5000Hz by using an acoustic impedance tube, and screening qualified products meeting the standard, thereby solving the technical problem that a gas film structure building in the prior art cannot realize high light transmittance and high sound absorption performance at the same time.

Description

Technical Field

[0001] The invention belongs to the technical field of building sound absorption and noise reduction, and in particular relates to a method for producing a light-transmitting sound-absorbing module suitable for an air membrane structure. Background Art

[0002] Air membrane structures are widely used in public buildings such as large venues and exhibition centers because of their lightness and good light transmittance. The acoustic treatment of traditional air membrane structure buildings mainly relies on porous sound-absorbing materials or micro-perforated panels for acoustic optimization to reduce indoor reverberation and improve speech clarity. Common technical methods include the use of suspended fiber sound absorbers, spray-on porous sound-absorbing coatings, or embedding sound-absorbing structures in air membrane substrates.

[0003] However, traditional air membrane structure acoustic treatment technology faces multiple defects: on the one hand, although conventional sound-absorbing materials such as glass wool and rock wool have good sound absorption performance, they seriously reduce the light transmittance of the air membrane structure and affect the natural lighting effect of the building; on the other hand, the micro-perforated sound-absorbing structure with high light transmittance has poor stability in complex environments, is easily affected by moisture and temperature changes, and the processing technology is complicated, making it difficult to apply to curved air membrane structures on a large scale.

[0004] Therefore, the field of air-supported membrane structures is in urgent need of a technical solution that can take into account both high light transmittance and high sound absorption performance. This has become a core technical problem that the industry needs to solve urgently. Summary of the invention

[0005] In view of this, the present invention provides a method for producing a light-transmitting and sound-absorbing module suitable for an air membrane structure, which can solve the technical problem in the prior art that air membrane structure buildings cannot achieve both high light transmittance and high sound absorption performance.

[0006] The present invention is implemented as follows: The present invention provides a method for producing a light-transmitting sound-absorbing module suitable for an air membrane structure, comprising: preparing a light-transmitting sound-absorbing substrate, selecting non-woven polyester fibers or glass fibers with high light transmittance, and forming a uniform fiber layer through a hot air web forming process; performing fireproofing treatment on the uniform fiber layer, preparing a flame retardant into a flame retardant liquid for immersion treatment and drying; placing the treated uniform fiber layer in a mold, and forming a plate-shaped sound absorber with a preset thickness distribution through a hot pressing molding process; using a MicroPore optimization algorithm to calculate micropore distribution parameters to output optimal micropore diameter parameters, micropore distribution density parameters, micropore shape coefficient parameters, and open porosity parameters, and based on the output parameters, covering a transparent microporous polymer film on the surface of the plate-shaped sound absorber, and compounding the transparent microporous polymer film with the plate-shaped sound absorber through an ultrasonic welding process; preparing and applying a waterproof and mildew-proof solution to treat a composite material; drying the composite material; applying a pre-trained AcousticDome neural network model to analyze the acoustic performance of the composite material and optimize the cutting plan; testing the standardized module sound absorption coefficient and screening qualified products.

[0007] Among them, in the step of preparing the light-transmitting sound-absorbing substrate, the selected non-woven polyester fiber or glass fiber has a light transmittance greater than 85%. A uniform fiber layer with a thickness of 30 to 50 millimeters is formed by the hot air forming process, and the density of the uniform fiber layer is controlled at 40 to 60 kilograms per cubic meter.

[0008] Among them, the hot air forming process is specifically to fully loosen the non-woven polyester fiber or glass fiber through a carding machine and then evenly spread it on a conveyor belt. Hot air with a temperature of 120 to 150 degrees Celsius is blown from the upper and lower directions simultaneously on the uniform fiber layer, so that the non-woven polyester fiber or glass fiber is entangled with each other to form a three-dimensional structure, enhancing the overall strength of the uniform fiber layer while maintaining a high porosity to improve the sound absorption performance.

[0009] Among them, in the step of fireproofing the uniform fiber layer, borate flame retardant is dissolved in deionized water to prepare a flame retardant solution with a mass fraction of 10% to 15%. The uniform fiber layer is immersed in the flame retardant solution for 5 to 10 minutes, and then dried at 70 to 80 degrees Celsius until the moisture content is lower than 2%.

[0010] Among them, in the step of the hot pressing forming process, it is pressed at 170 to 190 degrees Celsius and a pressure of 0.5 to 1.0 megapascals for 10 to 15 minutes to form a plate-shaped sound absorber with a preset thickness distribution.

[0011] Among them, the thickness of the transparent microporous polymer film is 0.05 to 0.1 millimeters.

[0012] Among them, the ultrasonic welding process is specifically to use high-frequency vibration energy to generate frictional heat of material molecules under a pressurized state, promoting a firm connection between the transparent microporous polymer film and the surface fibers of the plate-shaped sound absorber at the molecular level. No adhesive is used to avoid blocking the pores of the material and affecting the sound absorption performance. The ultrasonic frequency is 20 to 40 kilohertz.

[0013] Among them, in the step of preparing the waterproof and mildew-proof solution, the fluorocarbon water repellent and the quaternary ammonium salt mildew-proof agent are mixed at a mass ratio of 5:1, and an acrylic cross-linking agent is added to adjust the pH value to 6.5 to 7.5. The composite material of the transparent microporous polymer film and the plate-shaped sound absorber is immersed in the waterproof and mildew-proof solution for 30 to 60 minutes.

[0014] Among them, in the step of drying the composite material, it is dried at 60 to 70 degrees Celsius and a relative humidity of 30% to 40% for 120 to 180 minutes to ensure that the composite material is dried evenly and does not deform.

[0015] Among them, the MicroPore optimization algorithm is based on the MicroPore optimization algorithm function, which is used to calculate the optimal micropore distribution parameters according to the air film structure characteristics and acoustic requirements. The inputs include the acoustic characteristics of the air film structure, the sound absorption coefficient of the target frequency band, the required value of the light transmittance, the environmental temperature and humidity parameters, and the thickness of the plate-shaped sound absorber. The outputs are the optimal micropore diameter parameters, the micropore distribution density parameters, the micropore shape coefficient parameters, and the opening rate parameters. The MicroPore optimization algorithm function simultaneously considers the acoustic performance and the optical performance through a multi-objective optimization algorithm, and maximizes the sound absorption coefficient of the target frequency band on the premise of meeting the required value of the light transmittance.

[0016] Among them, in the steps of analyzing the acoustic performance of the composite material by applying the pre-trained AcousticDome neural network model, the sound absorption coefficients of each frequency band are predicted according to the thickness distribution of the composite material, the density gradient parameters, the optimal micropore diameter parameters, the micropore distribution density parameters, the micropore shape coefficient parameters, and the opening rate parameters. The specific structure of the AcousticDome neural network model is a hybrid architecture combining a multi-layer convolutional neural network and a graph attention network, which includes three main modules: a material physical property encoder, an acoustic response predictor, and a spatial configuration optimizer. Among them, the material physical property encoder uses a three-layer three-dimensional convolutional neural network to extract the structural characteristics of the composite material, the acoustic response predictor uses a four-layer self-attention mechanism to establish the interaction relationship between sound waves of different frequency bands and the composite material, and the spatial configuration optimizer considers the spatial geometric characteristics of the air film structure and the normalized module layout scheme through a graph neural network.

[0017] Compared with the prior art, for a method for producing a light-transmitting and sound-absorbing module applicable to an air film structure provided by the present invention, the method for producing the light-transmitting and sound-absorbing module proposed by the present invention successfully solves the technical problem that it is difficult to balance the light-transmitting and sound-absorbing performances in an air film structure building by selecting a high-light-transmitting fiber material, optimizing the micropore parameter design, applying a special waterproof and mildew-proof treatment, and an intelligent acoustic performance prediction technology.

[0018] While maintaining a light transmittance of more than 85%, the light-transmitting and sound-absorbing module produced by this method achieves an acoustic performance with a noise reduction coefficient NRC greater than 0.6 and a sound absorption coefficient in the middle frequency band (500 - 1000 Hz) greater than 0.7, which is significantly better than the traditional technology. By accurately calculating the micropore distribution parameters through the MicroPore optimization algorithm and combining the non-glue ultrasonic welding technology, the problem of pore blockage caused by the traditional adhesive process is avoided, effectively improving the synergistic effect of the sound absorption performance and the light transmittance. At the same time, the three-dimensional stereoscopic fiber structure formed by the hot air forming process and the waterproof and mildew-proof treatment greatly improve the stability of the material in the special environment of the air film.

[0019] Based on the acoustic performance prediction and module layout optimization of the AcousticDome neural network model, the present invention not only solves the performance problems of individual modules, but also can optimize the overall sound field according to the characteristics of different air film structures, solving the technical problem that it is difficult to balance the light transmission and sound absorption performance in air film structure buildings. Brief Description of the Drawings

[0020] Figure 1 It is a flowchart of the method of the present invention. Detailed Embodiments

[0021] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0022] As Figure 1 shown, it is a flowchart of a method for producing a light-transmitting and sound-absorbing module applicable to an air film structure provided by the present invention. The method includes the following steps:

[0023] S01. Prepare a light-transmitting and sound-absorbing substrate. Select non-woven polyester fiber or glass fiber with a light transmittance greater than 85%. Form a uniform fiber layer with a thickness of 30 to 50 millimeters from the non-woven polyester fiber or glass fiber through a hot air forming process, and control the density of the uniform fiber layer at 40 to 60 kilograms per cubic meter;

[0024] S02. Perform fireproof treatment on the uniform fiber layer. Dissolve borate flame retardant in deionized water to prepare a flame retardant solution with a mass fraction of 10% to 15%. Immerse the uniform fiber layer in the flame retardant solution for 5 to 10 minutes, and then dry it at 70 to 80 degrees Celsius until the moisture content is lower than 2%;

[0025] S03. Place the treated uniform fiber layer in a mold, and press it for 10 to 15 minutes at 170 to 190 degrees Celsius and a pressure of 0.5 to 1.0 megapascals through a hot pressing forming process to form a plate-shaped sound absorber with a preset thickness distribution;

[0026] S04. Use the MicroPore optimization algorithm to calculate the micropore distribution parameters. Input the acoustic characteristics of the air film structure, the sound absorption coefficient in the target frequency band, the required value of the light transmittance, the environmental temperature and humidity parameters, and the thickness of the plate-shaped sound absorber, and output the optimal micropore diameter parameter, the micropore distribution density parameter, the micropore shape coefficient parameter, and the opening ratio parameter. According to the optimal micropore diameter parameter, the micropore distribution density parameter, the micropore shape coefficient parameter, and the opening ratio parameter, cover a transparent microporous polymer film with a thickness of 0.05 to 0.1 millimeters on the surface of the plate-shaped sound absorber, and compound the transparent microporous polymer film with the plate-shaped sound absorber through an ultrasonic welding process at a frequency of 20 to 40 kilohertz;

[0027] S05. Prepare a waterproof and mildew-proof solution by mixing a fluorocarbon water repellent and a quaternary ammonium salt mildew-proof agent in a mass ratio of 5:1, adding an acrylic cross-linking agent to adjust the pH value to 6.5 to 7.5, and immersing the composite material of the transparent microporous polymer film and the plate-shaped sound absorber in the waterproof and mildew-proof solution for 30 to 60 minutes;

[0028] S06. Dry the treated composite material at 60 to 70 °C and a relative humidity of 30% to 40% for 120 to 180 minutes to ensure that the composite material is evenly dried and does not deform;

[0029] S07. Analyze the acoustic performance of the composite material using a pre-trained AcousticDome neural network model, predict the sound absorption coefficients of each frequency band according to the thickness distribution, density gradient parameters, the optimal micropore diameter parameters, the micropore distribution density parameters, the micropore shape coefficient parameters, and the porosity parameters of the composite material, optimize the cutting plan, cut the dried composite material into standardized modules according to the preset size, and perform a heat sealing process on the edges of the standardized modules to prevent internal fiber shedding and edge water absorption;

[0030] S08. Use an acoustic impedance tube to test the sound absorption coefficients of the standardized modules in the frequency band of 100 Hz to 5000 Hz according to national standards, and select qualified products with a noise reduction coefficient NRC greater than 0.6 and a sound absorption coefficient in the frequency band of 500 Hz to 1000 Hz greater than 0.7.

[0031] Among them, the hot air nonwoven process is specifically to fully loosen the nonwoven polyester fiber or glass fiber through a carding machine and then evenly spread it on a conveyor belt, and use hot air at a temperature of 120 to 150 °C to blow the uniform fiber layer from both the upper and lower directions at the same time, so that the nonwoven polyester fiber or glass fiber is entangled with each other to form a three-dimensional structure, enhancing the overall strength of the uniform fiber layer while maintaining a high porosity to improve the sound absorption performance.

[0032] Among them, the ultrasonic welding process is specifically to use high-frequency vibration energy to generate frictional heat in the material molecules under a pressurized state, so that the surface fibers of the transparent microporous polymer film and the plate-shaped sound absorber form a firm connection at the molecular level, and no adhesive is used to avoid blocking the pores of the material and affecting the sound absorption performance.

[0033] Among them, the heat sealing process is specifically to use a heat sealing knife at a temperature of 180 to 200 °C to press and melt along the edges of the standardized modules to form a sealed edge with a width of 5 to 10 mm, and at the same time form a transition structure with a gradually changing thickness in the edge area of the standardized modules to reduce the edge diffraction phenomenon of sound waves and improve the mid-high frequency sound absorption effect.

[0034] Among them, the acoustic impedance tube test specifically places the normalization module in an acoustic impedance tube with an inner diameter of 100 mm. A loudspeaker is installed at one end to generate plane sound waves, and the normalization module is placed at the other end. The sound absorption coefficient of the normalization module is determined by measuring the sound pressure at two points in the tube. During the test, the edge of the normalization module is sealed to avoid measurement errors caused by boundary effects.

[0035] The MicroPore optimization algorithm is based on the MicroPore optimization algorithm function and is used to calculate the optimal micropore distribution parameters according to the air film structure characteristics and acoustic requirements. The inputs include the acoustic characteristics of the air film structure, the sound absorption coefficient in the target frequency band, the required value of the light transmittance, the environmental temperature and humidity parameters, and the thickness of the plate-shaped sound absorber. The outputs are the optimal micropore diameter parameters, the micropore distribution density parameters, the micropore shape coefficient parameters, and the aperture ratio parameters. The MicroPore optimization algorithm function simultaneously considers the acoustic performance and optical performance through a multi-objective optimization algorithm, and maximizes the sound absorption coefficient in the target frequency band on the premise of meeting the required value of the light transmittance.

[0036] The specific structure of the AcousticDome neural network model is a hybrid architecture combining a multi-layer convolutional neural network and a graph attention network, including three main modules: a material physical property encoder, an acoustic response predictor, and a spatial configuration optimizer. Among them, the material physical property encoder uses a three-layer three-dimensional convolutional neural network to extract the composite material structure features. The acoustic response predictor uses a four-layer self-attention mechanism to establish the interaction relationship between sound waves in different frequency bands and the composite material. The spatial configuration optimizer considers the spatial geometric characteristics of the air film structure and the layout scheme of the normalization module through a graph neural network, and finally outputs the predicted value of the sound absorption coefficient in the frequency band and the optimal spatial layout suggestion.

[0037] The steps for establishing the training data set during the pre-training process of the AcousticDome neural network model specifically include collecting the measured sound absorption coefficient data under no less than 1000 different combinations of material parameters, including parameter changes such as the fiber material thickness from 10 to 100 mm, the density from 20 to 100 kg / m³, the micropore diameter from 0.1 to 5.0 mm, and the aperture ratio from 10% to 60%. At the same time, collect the acoustic environment parameters and reverberation time data of 50 typical air film structure buildings, use the Odeon sound field analysis software to simulate and verify the sound field characteristics, combine the on-site measurement data to construct a high-quality training sample set, standardize all samples, and divide them into a training set, a validation set, and a test set according to a ratio of 8:1:1.

[0038] The steps for pre-training the AcousticDome neural network model specifically include first initializing the network parameters with random weights, setting the learning rate to 0.001 using the Adam optimizer, with a batch size of 64, performing 300 rounds of iterative training on the training set, evaluating the performance of the AcousticDome neural network model on the validation set every 50 rounds, applying an early stopping strategy when the loss function on the validation set no longer decreases, then evaluating the generalization performance of the AcousticDome neural network model on the test set to ensure that the average prediction error is less than 5%, and then using transfer learning technology to fine-tune for the air film structure characteristics. During the fine-tuning process, the parameters of the material physical property encoder layer are frozen and only the parameters of the acoustic response predictor and the spatial configuration optimizer are updated. After 100 rounds of fine-tuning training, a model optimized for the air film structure is obtained.

[0039] The following describes the specific implementation manners of the above steps in detail.

[0040] The specific implementation manner of step S01 is to select non-woven polyester fiber or glass fiber with a light transmittance greater than 85% as the base material. Such materials have good light transmittance while maintaining a high sound absorption capacity. First, the selected fiber raw materials are transported to a carding machine for sufficient loosening treatment. The carding process uses carding-type loosening technology, and the fiber mass is fully broken up into single filaments by a high-speed rotating needle roller to ensure the uniformity of fiber distribution during the subsequent web-forming process. The loosened fibers are evenly spread on the conveyor belt through an air conveying system to form a preliminary loose fiber layer. Subsequently, the hot air system is started, and hot air with a temperature precisely controlled within the range of 120 to 150 degrees Celsius is blown on the fiber layer from both the upper and lower directions. The hot air flow rate is controlled at 1.5 to 2.5 meters per second, causing the fibers to entangle with each other under the action of the hot air flow. The hot air action causes partial melting on the fiber surface to form bonding points, and a three-dimensional network structure is formed through intermolecular van der Waals forces and mechanical entanglement, enhancing the overall strength of the fiber layer while maintaining a high porosity of 70% to 85%. Finally, a uniform fiber layer with a thickness of 30 to 50 millimeters and a density of 40 to 60 kilograms per cubic meter is formed. The pore structure of this fiber layer exhibits multi-scale distribution characteristics, which is beneficial to the absorption and attenuation of sound waves at different frequencies.

[0041] The specific implementation of step S02 is to conduct fireproof and flame-retardant treatment on the fiber layer to meet the fire safety requirements of the air-supported structure. First, prepare a borate flame retardant solution by mixing sodium tetraborate and boric acid in a mass ratio of 3:2 and dissolving them in deionized water to prepare a flame retardant solution with a mass fraction of 10% to 15%. The pH value of the solution is controlled within the range of 8.0 to 9.5. The temperature of the flame retardant solution is controlled at 25 to 35 degrees Celsius. Immerse the uniform fiber layer completely in the flame retardant solution, and control the immersion time at 5 to 10 minutes. During this period, turn the fiber layer over once every 1 minute to ensure that the flame retardant solution fully penetrates the internal structure of the fiber. After immersion, place the fiber layer in an adjustable-speed roller extrusion device and control the content of the residual flame retardant solution to be 120% to 150% of the dry weight of the fiber layer. Subsequently, place the fiber layer in a hot air circulation oven, control the temperature at 70 to 80 degrees Celsius, and control the relative humidity at 30% to 40% for drying. The drying time is adjusted according to the thickness of the fiber layer until the moisture content is lower than 2%. The mechanism of the flame retardant treatment is that the borate forms a vitrified protective layer on the surface of the fiber, melts at high temperature to generate a dense boron oxide barrier layer to prevent oxygen diffusion, and at the same time releases crystal water to absorb heat and reduce the temperature, inhibiting the progress of the combustion reaction chain.

[0042] The specific implementation of step S03 is to use a hot pressing process to prepare a plate-shaped sound absorber with a specific thickness distribution from the treated uniform fiber layer. First, customize a hot pressing mold according to the acoustic design requirements. The inner surface of the mold is treated with a Teflon coating to reduce adhesion. The cavity of the mold is designed as a structure with a gradually changing thickness. The thickness of the central area is designed to be 25 to 35 millimeters, and the thickness of the edge area is designed to be 15 to 25 millimeters, forming a sound wave guiding type thickness gradient. Place the flame retardant treated uniform fiber layer into the mold and use a hydraulic hot press for molding. Control the pressing temperature at 170 to 190 degrees Celsius. This temperature range makes the surface of the fiber reach the softening point but not completely melt. Control the pressure at 0.5 to 1.0 MPa, and the pressing time is 10 to 15 minutes. The pressing process adopts a stepped pressure increase method. In the initial stage, increase the pressure at a rate of 0.1 MPa per minute to 0.3 MPa and hold for 3 minutes to fully preheat and soften the fiber; then increase the pressure to the set pressure at a rate of 0.2 MPa per minute and hold until the molding is completed. During the hot pressing process, the fiber material undergoes local melting and bonding under the dual action of heat energy and pressure, forming a stable three-dimensional network structure. The density distribution shows the characteristic of low in the center and high at the edge. This gradient density distribution is beneficial to the damping attenuation and energy conversion of sound waves inside the material. After pressing, slowly cool it to below 80 degrees Celsius before demolding to avoid deformation caused by thermal stress.

[0043] The specific implementation of step S04 is to use the MicroPore optimization algorithm to calculate and apply the microporous membrane structure to improve the acoustic performance of the composite material. The MicroPore optimization algorithm is constructed based on the acoustic transfer matrix theory and the optical transmittance model, and the correlation between acoustic performance and optical performance is established through multi-parameter coupling. First, input parameters are collected, including the acoustic characteristics of the air film structure (including parameters such as reflection coefficient, absorption coefficient, and transmission coefficient), the target band absorption coefficient (usually setting target values for five frequency points of 250Hz, 500Hz, 1000Hz, 2000Hz, and 4000Hz), the required transmittance value (usually in the range of 75% to 85%), the environmental temperature and humidity parameters (including three parameters of temperature, humidity, and air pressure), and the thickness distribution data of the plate-shaped sound absorber. The algorithm uses a multi-objective genetic algorithm framework, with the population size set to 200, the number of evolutionary generations set to 500 generations, the crossover probability to 0.85, and the mutation probability to 0.15. The best solution is selected through the Pareto optimal set. The output parameters include the optimal micropore diameter parameter (in the range of 0.2 to 2.0 mm), the micropore distribution density parameter (in the range of 50 to 350 per square centimeter), the micropore shape factor parameter (shape parameters for circular, elliptical, or polygonal shapes), and the porosity parameter (in the range of 15% to 45%). According to the calculation results, a transparent polyester or polypropylene microporous film with a thickness of 0.05 to 0.1 mm is selected, and the microporous structure is processed according to the design parameters using precision laser drilling technology. The microporous film is covered on the surface of the plate-shaped sound absorber, and the two are compounded using ultrasonic welding technology. The output frequency of the ultrasonic generator is set at 20 to 40 kHz, the welding power is controlled at 500 to 800 W, the welding pressure is controlled at 0.2 to 0.4 MPa, and the welding time is 0.5 to 2.0 s. Ultrasonic welding uses high-frequency mechanical vibration to generate local frictional heat at the material interface, enabling the material molecules to diffuse and crosslink with each other under pressure, forming a firm connection without using adhesives, thus avoiding blocking the material pores and affecting the sound absorption performance.

[0044] The specific implementation of step S05 is to perform waterproof and mildew-proof treatment on the composite material to improve its service life in the air film structure environment. First, prepare a waterproof and mildew-proof solution. Select perfluorotridecyltriethoxysilane as the fluorocarbon water repellent and alkyl quaternary ammonium salt as the mildew-proof agent, and mix them in a mass ratio of 5:1. Add an acrylic crosslinking agent (glycidyl methacrylate) to the mixture to adjust the pH value of the solution to the range of 6.5 to 7.5. The addition amount of the crosslinking agent is controlled at 2% to 5% of the total mass, and the solid content of the solution is controlled at 8% to 12%. Immerse the composite material of the transparent microporous polymer film and the plate-shaped sound absorber completely in the waterproof and mildew-proof solution. The solution temperature is controlled at 20 to 30 degrees Celsius, and the soaking treatment time is 30 to 60 minutes. During this period, the material is flipped every 10 minutes to ensure uniform penetration of the solution. The waterproof mechanism is that the long-chain fluorocarbon groups in the fluorocarbon water repellent molecules form a low surface energy layer on the material surface, and the water droplet contact angle reaches more than 120 degrees, presenting superhydrophobicity; the mildew-proof mechanism is that the positive charge part in the quaternary ammonium salt molecules destroys the cell membrane structure of the mold and inhibits cell respiration and energy metabolism. The acrylic crosslinking agent forms a three-dimensional network structure on the fiber surface through free radical polymerization reaction, enhancing the bonding force and durability between the waterproof and mildew-proof layer and the substrate.

[0045] The specific implementation of step S06 is to perform a precisely controlled drying process on the treated composite material to ensure stable material performance and no deformation in shape. Place the composite material treated with waterproof and mildew-proof in a drying chamber with controllable temperature and humidity. The temperature is precisely controlled at 60 to 70 degrees Celsius, and the relative humidity is controlled at 30% to 40% for drying. The drying time is 120 to 180 minutes. The drying process adopts a staged control strategy. In the first stage, the temperature is set at 60 degrees Celsius and the relative humidity is 40%, lasting for 30 to 45 minutes, which is used to slowly evaporate the water on the material surface; in the second stage, the temperature rises to 65 degrees Celsius and the relative humidity drops to 35%, lasting for 60 to 90 minutes, which is used to evaporate the water inside the material; in the third stage, the temperature rises to 70 degrees Celsius and the relative humidity drops to 30%, lasting for 30 to 45 minutes, which is used to completely remove the residual water and ensure that the moisture content of the material is lower than 1%. During the drying process, the material is laid flat on a porous and breathable support frame and flipped every 30 minutes to prevent warping deformation caused by one-sided drying. A uniform air supply system is set in the drying chamber, and the wind speed is controlled at 0.5 to 1.2 meters per second to ensure the uniformity of the drying environment. After drying, the material is naturally cooled to room temperature and subjected to a 24-hour performance stabilization static treatment to fully release the internal stress of the material and avoid deformation during later use.

[0046] The specific implementation of step S07 is to analyze the acoustic properties of composite materials and perform modular processing using the pre-trained AcousticDome neural network model. First, import the composite material parameters, including thickness distribution data, density gradient parameters, optimal micropore diameter parameters, micropore distribution density parameters, micropore shape coefficient parameters, and porosity parameters, etc. Use the AcousticDome neural network model to predict the sound absorption coefficient curve of the material in the frequency range of 100 Hz to 5000 Hz, with a prediction resolution of 1 / 3 octave. In the model calculation process, use a convolutional neural network to extract the material structure features and establish an acoustic response mapping relationship. Based on the prediction results, apply a sound field uniformity optimization algorithm to generate the best cutting plan. The algorithm considers the spatial geometric characteristics of the air film structure and the sound field distribution characteristics, and cuts the dried composite material into standardized modules according to the preset size. Common specifications include 600 mm × 600 mm, 600 mm × 1200 mm, or custom sizes. The cutting uses ultrasonic cutting technology, with a cutting frequency of 40 to 60 kHz and a cutting speed controlled at 10 to 20 meters per minute. During the cutting process, the edge is simultaneously melted and sealed to prevent fiber shedding. Perform heat-sealing treatment on the edges of the cut standardized modules. Use a heat-sealing machine to set the temperature to 180 to 200 °C, the pressure to 0.3 to 0.6 MPa, and the heat-sealing time to 3 to 8 seconds to form a sealed edge with a width of 5 to 10 mm. The heat-sealing process melts the fibers and the microporous membrane in the edge area to form an integral structure, and at the same time forms a transition structure with a gradually changing thickness in the edge area of the module. The thickness gradually decreases from the center to the edge, with a reduction rate of 15% to 25%. This structure reduces the edge diffraction phenomenon of sound waves and improves the mid-high frequency sound absorption effect.

[0047] The specific implementation of step S08 is to conduct standardized acoustic performance tests on the normalization module and select qualified products for use. The test uses the acoustic impedance tube method that complies with national standards. The inner diameter of the acoustic impedance tube is 100 mm, which is applicable to measuring the sound absorption coefficient in the frequency band of 100 Hz to 5000 Hz. The test equipment includes a precision sound source, a dual-microphone system, and a data acquisition and analysis system. The sound pressure level generated by the sound source is in the range of 90 to 100 dB, the sampling frequency is not less than 44.1 kHz, the spectral analysis uses the fast Fourier transform algorithm, and the frequency resolution is not less than 2 Hz. Place the normalization module sample at the end of the acoustic impedance tube, and use sound-impermeable tape to seal the gap between the sample edge and the tube wall to avoid sound leakage. During the test, collect the sound pressure data at two measurement points inside the tube, calculate the complex reflection coefficient of the sample through the transfer function method, and then obtain the sound absorption coefficient in each frequency band. The noise reduction coefficient NRC is obtained by calculating the arithmetic mean of the sound absorption coefficients at four frequency points of 250 Hz, 500 Hz, 1000 Hz, and 2000 Hz. Select products with an NRC greater than 0.6 and a sound absorption coefficient in the frequency band of 500 Hz to 1000 Hz greater than 0.7 as qualified products. During the test process, the ambient temperature is controlled at 20 ± 2 °C, the relative humidity is controlled at 50 ± 5%, the atmospheric pressure is recorded and used for the calculation of sound speed correction, and each sample is tested 3 times and the average value is taken as the final result.

[0048] The specific implementation of the AcousticDome neural network model is to design a hybrid architecture neural network for predicting and optimizing the acoustic performance of the light-transmitting and sound-absorbing modules in the air film structure. The model is based on an architecture that combines a multi-layer convolutional neural network and a graph attention network, and includes three core modules: a material physical property encoder, an acoustic response predictor, and a spatial configuration optimizer. The material physical property encoder adopts a three-layer three-dimensional convolutional neural network structure. The first layer uses 64 3×3×3 convolutional kernels, with a stride of 1, padding of 1, and an activation function of ReLU; the second layer uses 128 3×3×3 convolutional kernels, with a stride of 2, padding of 1, and an activation function of ReLU; the third layer uses 256 3×3×3 convolutional kernels, with a stride of 2, padding of 1, and an activation function of ReLU, extracting the three-dimensional structure features of the material and reducing the dimension to a feature vector. The acoustic response predictor adopts a four-layer self-attention mechanism. Each layer contains 8 attention heads, the hidden layer dimension is 512, the feed-forward neural network dimension is 2048, the activation function is GELU, and residual connections and layer normalization are applied to each layer to establish the interaction relationship between sound waves in different frequency bands and the material structure. The spatial configuration optimizer is implemented based on a graph neural network, adopting an alternating stacking structure of graph convolutional layers and graph attention layers. The graph convolutional layer is implemented using the ChebNet algorithm, and the graph attention layer is implemented using the GAT algorithm, considering the spatial geometric characteristics of the air film structure and the module layout scheme. The final output layer uses a fully connected layer to map the features to the predicted values of the sound absorption coefficients in each frequency band and the spatial layout suggestions. The loss function combines the mean square error and the cross-entropy loss, and adjusts the relative importance of the two through weight coefficients.

[0049] The specific implementation method for establishing the training dataset is to construct a high-quality acoustic performance dataset for the training and optimization of the AcousticDome neural network model. First, collect the measured sound absorption coefficient data under no less than 1000 different combinations of material parameters. The parameter coverage range includes variations such as the fiber material thickness from 10 to 100 mm, density from 20 to 100 kg / m³, fiber diameter from 5 to 30 µm, fiber length from 20 to 80 mm, micropore diameter from 0.1 to 5.0 mm, porosity from 10% to 60%, and micropore shape from circular to polygonal, etc., to ensure that the data distribution is uniform and representative. Obtain the sound absorption coefficient curve of the standard sample in the frequency range of 100 Hz to 5000 Hz through laboratory acoustic impedance tube testing, with a frequency resolution of not less than 1 / 3 octave. At the same time, collect the acoustic environment parameters of 50 typical air-supported structure buildings, including indoor reverberation time (RT60), background noise level (NC value), speech intelligibility index (STI value), sound field uniformity and other parameters, and the collection points cover the key positions of the entire spatial volume. Use Odeon sound field analysis software to establish a digital model of the air-supported structure, and simulate and verify the sound field characteristics. The software parameter settings include the number of rays not less than 10000, the reflection order not less than 2000, the scattering coefficient taking the measured value or a reasonable estimated value, and the sound absorption coefficient using the measured data. Compare and verify the measured data with the simulation results, and require the error of the two in the key acoustic parameters to be controlled within ±10% to ensure the accuracy of the model. Perform standardized preprocessing on all sample data, including outlier detection and removal, missing value imputation, feature scaling and normalization, and divide the data into a training set, a validation set, and a test set according to the ratio of 8:1:1.

[0050] Optionally, the transparent microporous polymer film is prepared from a high-transparency polyester material (PET) or polypropylene (PP). The film thickness is precisely controlled within the range of 0.05 to 0.1 mm, with a light transmittance of not less than 95% and a haze of less than 2%. The molecular weight of the high-transparency polyester material is controlled within the range of 25,000 to 35,000 Daltons, and the crystallinity is controlled within the range of 5% to 15% to ensure good light transmittance and mechanical strength of the film. The preparation process uses the biaxial stretching method. First, the polymer raw material is dried (at 80 °C, in a vacuum environment, for 4 to 6 hours), and the residual moisture is controlled below 50 ppm. Then, it is melt-extruded into an initial film at 275 to 285 °C. After cooling to 90 °C, under the condition of a longitudinal stretching temperature of 115 °C, the stretching ratio is 3.5 to 4.0. Subsequently, under the condition of a transverse stretching temperature of 125 °C, the stretching ratio is 3.8 to 4.2. Finally, heat setting treatment is carried out (temperature 205 °C, time 15 to 20 seconds). The microporous structure is processed using precision laser drilling technology. An ultraviolet laser with a wavelength of 355 nm is used, the pulse width is controlled within the range of 12 to 15 nanoseconds, the frequency is 50 to 65 kHz, and the single-pulse energy density is 3.5 to 4.5 joules per square centimeter. According to the calculation results of the MicroPore optimization algorithm, the micropore diameter is controlled within the range of 0.2 to 2.0 mm, the micropore distribution density is 50 to 350 per square centimeter, the micropore shape can be circular, elliptical or polygonal, and the aperture ratio is controlled within the range of 15% to 45%. The film after laser drilling is subjected to plasma surface treatment to improve hydrophilicity. The treatment gas is a mixture of oxygen and argon (ratio 1:4), the power density is 25 to 35 watts per square centimeter, and the treatment time is 10 to 20 seconds, so that the surface energy of the film is increased to 42 to 48 dynes per centimeter, which is beneficial to subsequent ultrasonic welding and waterproof and mildew-proof treatment.

[0051] The following details the mathematical models or calculation processes involved in the present invention.

[0052] The MicroPore optimization algorithm in step S04 involves multiple core calculation processes, and this algorithm is constructed based on the acoustic transfer matrix theory and the optical transmittance model. The acoustic performance calculation is specifically expressed as follows:

[0053]

[0054] In the formula, Z c is the acoustic impedance of the micropore; ρ0 is the air density, with a value range of 1.18~1.25 kg / m 3 ; c0 is the speed of sound, with a value range of 340~345 m / s; η is the air dynamic viscosity coefficient, with a value range of 1.75×10 -5 ~1.85×10 -5Pa·s; ω is the angular frequency, ω = 2πf, where f is the acoustic wave frequency, ranging from 100 to 5000 Hz; d is the micropore diameter, ranging from 0.2 to 2.0 mm; i is the imaginary unit.

[0055] This formula is based on the Maa microporous plate theory, taking into account the viscous effect and heat conduction effect of the gas in the micropores. By introducing the imaginary unit to handle the phase change of the acoustic wave, it can more accurately characterize the acoustic impedance characteristics of the microporous structure compared with the traditional model.

[0056] The total acoustic impedance of the microporous composite material is calculated as follows:

[0057]

[0058] In the formula, Z t is the total acoustic impedance of the composite material; Z f is the acoustic impedance of the fiber layer; k is the wave number, k = ω / c0; d f is the thickness of the fiber layer, ranging from 25 to 50 mm.

[0059] The acoustic impedance Z f of the fiber layer is calculated through the Delany - Bazley model:

[0060]

[0061] In the formula, ρ f is the density of the fiber layer, ranging from 40 to 60 kg / m 3 ; σ is the flow resistance of the fiber layer,

[0062] The flow resistance σ is obtained through experimental measurement. The specific steps include: placing the sample in a standard flow resistance test device, controlling the air flow rate in the range of 0.5 - 5.0 mm / s, measuring the pressure difference Δp and volume flow rate q on both sides of the sample, and calculating through σ = Δp·A / (q·d f ), where A is the cross - sectional area of the sample.

[0063] The sound absorption coefficient of the material is calculated as follows:

[0064]

[0065] In the formula, α(ω) is the sound absorption coefficient at a specific angular frequency ω, ranging from 0 to 1.

[0066] The optical transmittance model is calculated as follows:

[0067] T r =(1 - β)·(1 - γ·p);

[0068] In the formula, T ris the total light transmittance of the composite material; β is the light attenuation coefficient of the fiber layer, with a range of 0.05 to 0.15; γ is the light scattering coefficient of the micropores, with a range of 0.02 to 0.08; p is the porosity, with a range of 0.15 to 0.45.

[0069] The parameter β is obtained through experimental measurement. Samples of fiber layers with different thicknesses and densities are placed in a standard integrating sphere light transmittance tester to measure the incident light intensity I0 and the transmitted light intensity I t , and it is calculated by β = 1 - I t / I0. The parameter γ is obtained by measuring a microporous membrane sample through a similar method.

[0070] The multi-objective optimization problem is expressed as:

[0071] minF(x) = (f1(x), f2(x));

[0072]

[0073] f2(x) = -(T r -T min );

[0074] In the formula, F(x) is the objective function vector; f1(x) is the acoustic performance objective function, taking a negative value to convert the optimization problem into a minimization problem; f2(x) is the optical performance objective function; w i is the frequency weight coefficient, α(f i ) is the sound absorption coefficient at the frequency f i ; T min is the minimum light transmittance requirement value, usually 0.75 to 0.85; x is the design variable vector, x = (d, p, s, ρ f , d f ), including the micropore diameter d, the porosity p, the micropore shape coefficient s, the fiber density ρ f and the fiber thickness d f .

[0075] The constraint conditions of the design variables are:

[0076] 0.2 ≤ d ≤ 2.0;

[0077] 0.15 ≤ p ≤ 0.45;

[0078] 1 ≤ s ≤ 3;

[0079] 40 ≤ ρ f ≤ 60;

[0080] 25 ≤ d f ≤ 50;

[0081] The multi-objective optimization is solved using the NSGA-II algorithm, with a population size of 200, an evolutionary generation of 500, a crossover probability of 0.85, and a mutation probability of 0.15. This algorithm constructs the Pareto optimal set through non-dominated sorting and crowding distance calculation, and selects the best solution that meets the light transmittance requirements from it.

[0082] In the AcousticDome neural network model, the material physical property encoder uses a three-dimensional convolutional neural network, and the convolution operation is expressed as:

[0083]

[0084] In the formula, is the value of the k-th channel of the output feature map at the position (l, m, n); is the value of the i-th channel of the input feature map at the position (l + p, m + q, n + r); is the weight of the convolution kernel from the input channel i to the output channel k at the position (p, q, r); b k is the bias term of the k-th output channel; F i is the number of input channels; F h 、F w 、F d are the height, width, and depth of the convolution kernel respectively.

[0085] The convolution layer is followed by the ReLU activation function:

[0086] f(x) = max(0, x);

[0087] Optionally, the self-attention mechanism in the acoustic response predictor is calculated as follows:

[0088] Q = XW Q ;

[0089] K = XW K ;

[0090] V = XW V ;

[0091]

[0092] Z = AV;

[0093] In the formula, X is the input feature matrix; W Q 、W K 、W V are the weight matrices of the query, key, and value; d k is the dimension of the key vector; A is the attention weight matrix; Z is the attention output.

[0094] Optionally, the software function softmax is expressed as:

[0095] Optionally,

[0096] The multi-head attention mechanism concatenates the outputs of individual attentions:

[0097] MultiHead(X) = Concat(Z1, Z2, ..., Z h )W O ;

[0098] where Z1, Z2, ..., Z h are the outputs of h attention heads; W O is the output projection matrix; h is the number of attention heads, taking the value of 8.

[0099] Optionally, the graph convolution operation in the spatial configuration optimizer is expressed as:

[0100]

[0101] where H (l) is the node feature matrix of the l-th layer; is the adjacency matrix with self-connection added; is 's degree matrix; W (l) is the weight matrix of the l-th layer; σ is the activation function.

[0102] Optionally, the graph attention layer is calculated as follows:

[0103]

[0104] where is the feature vector of node i at the l-th layer; a is the attention coefficient calculation function; is the neighbor set of node i; α ij is the attention weight of node j to node i.

[0105] Optionally, the loss function of the model combines the mean squared error and the cross-entropy loss:

[0106]

[0107] where y i is the actual sound absorption coefficient; is the predicted sound absorption coefficient; z j is the actual layout label; is the predicted layout probability; λ1, λ2 are weight coefficients, taking the values of 0.7 and 0.3 respectively; n is the number of sound absorption coefficient samples; m is the number of layout samples.

[0108] These equations together constitute the theoretical basis for the design and optimization of the light-transmitting and sound-absorbing module, and solve the balance problem between acoustic performance and optical performance in the air film structure through a multidisciplinary coupling method. The MicroPore optimization algorithm accurately predicts the sound-absorbing performance of materials through the acoustic transfer matrix theory, and realizes multi-objective optimization while considering the optical transmittance constraint.

[0109] The AcousticDome neural network model integrates deep learning and acoustic physical models. It extracts the material structure features through a convolutional neural network, establishes a frequency-response mapping through a self-attention mechanism, and optimizes the spatial layout through a graph neural network, forming a complete design and optimization system for the light-transmitting and sound-absorbing module.

[0110] Specifically, the principle of the present invention is as follows: The technical principle of the present invention is based on the collaborative optimization of the acoustic theory of porous materials and the optical transmission principle. Through the integrated application of material structure design, precise control of the microstructure, and advanced artificial intelligence technology, the balance between light transmission and sound absorption performance is achieved.

[0111] First, a non-woven polyester fiber or glass fiber with a light transmittance greater than 85% is selected as the base material, and a three-dimensional structure is formed by the hot air nonwoven process, which improves the overall light transmittance of the material while ensuring a porous structure. The voids between the fibers formed by the hot air nonwoven process provide rich propagation channels and frictional attenuation interfaces for sound waves, enabling the sound energy to be converted into heat energy through viscous loss and heat exchange. By controlling the fiber layer density in the range of 40 to 60 kilograms per cubic meter, an appropriate flow resistance is formed to optimize the attenuation efficiency of medium and low-frequency sound wave energy.

[0112] Secondly, the MicroPore optimization algorithm realizes multi-objective optimization of acoustic performance and optical performance by accurately calculating the micropore distribution parameters. The algorithm is based on an acoustic theory model, considering the influence of micropore diameter, distribution density, shape factor, and porosity on the acoustic impedance, and maximizing the sound absorption coefficient in a specific frequency band on the premise of meeting the light transmission requirements. The thin-film micropore structure and the base material fiber layer form a "double-layer composite resonance system", which uses the micropore resonance sound absorption principle to enhance the sound absorption performance in the medium and high-frequency bands, while the fiber base material enhances the attenuation of sound energy in the low-frequency band.

[0113] Thirdly, ultrasonic welding technology is used to achieve glue-free compounding, avoiding the influence of traditional adhesives blocking pores on the sound absorption performance. At the same time, waterproof and mildew-proof treatment ensures the stable performance of the material in a high-humidity environment. The AcousticDome neural network model extracts the material structure features through a material physical property encoder, establishes the interaction relationship between sound waves and materials through an acoustic response predictor, and considers the geometric characteristics of the air film through a spatial configuration optimizer, realizing the full-scale optimization from the microstructure to the macroscopic sound field.

[0114] Finally, the thickness-gradual transition structure formed by the heat-sealing edge process reduces the edge diffraction phenomenon of sound waves, improves the mid-high frequency sound absorption effect, and enables the entire light-transmitting sound-absorbing module to stably and efficiently perform its functions in the air film structure environment.

[0115] A specific Embodiment 1 of the present invention is provided below. The specific implementation manners of each step in this Embodiment 1 are described in detail as follows.

[0116] The specific implementation manners of steps S01 - S03 are the same as those described above, and will not be elaborated here.

[0117] The specific implementation manner of step S04 is to calculate and apply a microporous membrane structure using the MicroPore optimization algorithm to enhance the acoustic performance of the composite material. The MicroPore optimization algorithm is constructed based on the acoustic transfer matrix theory and the optical transmittance model, and the correlation between acoustic performance and optical performance is established through multi-parameter coupling. First, input parameters are collected, including the acoustic characteristics of the air film structure, the sound absorption coefficient in the target frequency band, the required value of light transmittance, the environmental temperature and humidity parameters, and the thickness of the plate-shaped sound absorber. The acoustic performance is calculated using the microporous acoustic impedance model: In the formula, Z c is the microporous acoustic impedance; ρ0 is the air density, with a value range of 1.18 - 1.25 kg / m 3 ; c0 is the speed of sound, with a value range of 340 - 345 m / s; η is the air dynamic viscosity coefficient, with a value range of 1.75×10 -5 -1.85×10 -5 Pa·s; ω is the angular frequency, ω = 2πf, f is the sound wave frequency, in the range of 100 - 5000 Hz; d is the micropore diameter, in the range of 0.2 - 2.0 mm; i is the imaginary unit. The total acoustic impedance of the microporous composite material is calculated as: In the formula, Z t is the total acoustic impedance of the composite material; Z f is the acoustic impedance of the fiber layer; k is the wave number, k = ω / c0; d f is the thickness of the fiber layer, in the range of 25 - 50 mm. The acoustic impedance of the fiber layer is calculated through the Delany - Bazley model: In the formula, ρ f is the density of the fiber layer, in the range of 40 - 60 kg / m 3 ; σ is the flow resistance of the fiber layer, The sound absorption coefficient of the material is calculated as: In the formula, α(ω) is the sound absorption coefficient at a specific angular frequency ω. The optical transmittance model is: T r =(1 - β)·(1 - γ·p); in the formula, T ris the total light transmittance of the composite material; β is the light attenuation coefficient of the fiber layer, with a range of 0.05 - 0.15; γ is the light scattering coefficient of the micropores, with a range of 0.02 - 0.08; p is the porosity, with a range of 0.15 - 0.45. The multi-objective optimization problem is expressed as: minF(x) = (f1(x), f2(x)); f2(x) = -(T r -T min ); where F(x) is the objective function vector; f1(x) is the acoustic performance objective function; f2(x) is the optical performance objective function; w i is the frequency weight coefficient; α(f i ) is the sound absorption coefficient at frequency f i ; T min is the minimum light transmittance requirement value; x is the design variable vector. This optimization problem is solved by the non-dominated sorting genetic algorithm (NSGA-II) to obtain the optimal micropore distribution parameters. According to the calculation results, a transparent polyester or polypropylene microporous film with a thickness of 0.05 to 0.1 mm is selected, and the micropore structure is processed according to the design parameters using precision laser drilling technology, and is covered on the surface of the plate-shaped sound absorber. The two are compounded by ultrasonic welding process under the conditions of a frequency of 20 to 40 kHz, a power of 500 to 800 W, and a pressure of 0.2 to 0.4 MPa, and the welding time is 0.5 to 2.0 seconds.

[0118] The specific implementation manners of steps S05 - S06 are the same as those described above and will not be elaborated here.

[0119] The specific implementation manner of step S07 is to analyze the acoustic performance of the composite material using the pre-trained AcousticDome neural network model and perform modular processing. First, import the composite material parameters, including thickness distribution data, density gradient parameters, optimal micropore diameter parameters, micropore distribution density parameters, micropore shape coefficient parameters, and porosity parameters, etc. The material physical property encoder in the AcousticDome neural network model uses a three-dimensional convolutional neural network, and the convolution operation is expressed as: where is the value of the k-th channel of the output feature map at position (l, m, n); is the value of the i-th channel of the input feature map at position (l + p, m + q, n + r); is the weight of the convolution kernel from input channel i to output channel k at position (p, q, r); b k is the bias term of the k-th output channel; F i is the number of input channels; F h , F w , F dThey are the height, width, and depth of the convolutional kernel respectively. The self-attention mechanism in the acoustic response predictor is calculated as follows: Q = XW Q ; K = XW K ; V = XW V ; Z = AV; where X is the input feature matrix; W Q , W K , W V are the weight matrices for query, key, and value respectively; d k is the dimension of the key vector; A is the attention weight matrix; Z is the attention output. The graph convolution operation in the spatial configuration optimizer is expressed as: where H (l) is the node feature matrix of the l-th layer; is the adjacency matrix with self-connection added; is 's degree matrix; W (l) is the weight matrix of the l-th layer; σ is the activation function. The loss function of the model combines mean squared error and cross-entropy loss: where y i is the actual sound absorption coefficient; is the predicted sound absorption coefficient; z j is the actual layout label; is the predicted layout probability; λ1 and λ2 are weight coefficients, taking values 0.7 and 0.3 respectively. The sound absorption coefficient curve of the material in the frequency range of 100 Hz to 5000 Hz is predicted by this model, and then the acoustic field uniformity optimization algorithm is applied to generate the best cutting plan. The dried composite material is cut into standardized modules according to the preset size. The common sizes include 600 mm × 600 mm, 600 mm × 1200 mm, or custom sizes. The cutting uses ultrasonic cutting technology, with a cutting frequency of 40 to 60 kHz and a cutting speed controlled at 10 to 20 meters per minute. The cutting process simultaneously realizes edge melting and sealing to prevent fiber shedding. The edges of the cut standardized modules are subjected to heat-sealing treatment. The temperature of the heat-sealing machine is set at 180 to 200 °C, the pressure is set at 0.3 to 0.6 MPa, and the heat-sealing time is 3 to 8 seconds, forming a sealed edge with a width of 5 to 10 mm. The heat-sealing process melts the fibers and the microporous membrane in the edge area to form an integral structure, and at the same time forms a transition structure with a gradually changing thickness in the edge area of the module. The thickness gradually decreases from the center to the edge, with a reduction rate of 15% to 25%. This structure reduces the edge diffraction phenomenon of sound waves and improves the mid-high frequency sound absorption effect.

[0120] The specific implementation of step S08 is the same as the foregoing and will not be elaborated here

[0121] The specific implementation of the AcousticDome neural network model is to design a hybrid architecture neural network for predicting and optimizing the acoustic performance of the light-transmitting and sound-absorbing modules in the air film structure. This model is based on an architecture that combines a multi-layer convolutional neural network and a graph attention network, and includes three core modules: a material physical property encoder, an acoustic response predictor, and a spatial configuration optimizer. The material physical property encoder adopts a three-layer three-dimensional convolutional neural network structure. The first layer uses 64 3×3×3 convolutional kernels, with a stride of 1, padding of 1, and the activation function is ReLU, expressed as: f(x) = max(0, x); the second layer uses 128 3×3×3 convolutional kernels, with a stride of 2, padding of 1, and the activation function is ReLU; the third layer uses 256 3×3×3 convolutional kernels, with a stride of 2, padding of 1, and the activation function is ReLU, extracting the three-dimensional structure features of the material and reducing the dimension to a feature vector. The acoustic response predictor adopts a four-layer self-attention mechanism. Each layer contains 8 attention heads, the hidden layer dimension is 512, the feed-forward neural network dimension is 2048, the activation function is GELU, and residual connections and layer normalization are applied to each layer to establish the interaction relationship between sound waves of different frequency bands and the material structure. The multi-head attention mechanism concatenates the outputs of individual attention: MultiHead(X) = Concat(Z1, Z2,..., Z h )W O ; where Z1, Z2,..., Z h are the outputs of h attention heads; W O is the output projection matrix; h is the number of attention heads, and the value is 8. The spatial configuration optimizer is implemented based on a graph neural network, adopting an alternating stacking structure of graph convolutional layers and graph attention layers. The graph convolutional layer is implemented using the ChebNet algorithm, and the graph attention layer is implemented using the GAT algorithm. The calculation of the graph attention layer is as follows:

[0122] where, is the feature vector of node i at the l-th layer; a is the attention coefficient calculation function; is the neighbor set of node i; α ij is the attention weight of node j to node i. The final output layer uses a fully connected layer to map the features to the predicted values of the sound absorption coefficients in each frequency band and the spatial layout suggestions, comprehensively considering the spatial geometric characteristics of the air film structure and the layout scheme of the light-transmitting and sound-absorbing modules, and realizing the comprehensive optimization of acoustic performance and spatial aesthetics.

[0123] The specific implementation of establishing the training dataset is to construct a high-quality acoustic performance dataset for the training and optimization of the AcousticDome neural network model. First, collect the measured sound absorption coefficient data under no less than 1000 different combinations of material parameters. The parameter coverage range includes changes such as the fiber material thickness from 10 to 100 mm, density from 20 to 100 kg / m³, fiber diameter from 5 to 30 μm, fiber length from 20 to 80 mm, micropore diameter from 0.1 to 5.0 mm, porosity from 10% to 60%, and micropore shape from circular to polygonal, ensuring that the data distribution is uniform and representative. Obtain the sound absorption coefficient curve of the standard sample in the frequency band from 100 Hz to 5000 Hz through laboratory acoustic impedance tube testing, with a frequency resolution of not less than 1 / 3 octave. At the same time, collect the acoustic environment parameters of 50 typical air-supported structure buildings, including indoor reverberation time (RT60), background noise level (NC value), speech intelligibility index (STI value), sound field uniformity and other parameters, and the collection points cover the key positions of the entire spatial volume. Use Odeon sound field analysis software to establish a digital model of the air-supported structure, simulate and verify the sound field characteristics. The software parameter settings include the number of rays not less than 10000, the reflection order not less than 2000, the scattering coefficient taking the measured value or a reasonable estimated value, and the sound absorption coefficient using the measured data. Compare and verify the measured data with the simulation results, and require the error of the two in key acoustic parameters to be controlled within ±10% to ensure the accuracy of the model. Perform standardized preprocessing on all sample data, including outlier detection and removal, missing value imputation, feature scaling and normalization, and divide the data into a training set, a validation set and a test set according to the ratio of 8:1:1. The training process first initializes the network parameters with random weights, uses the Adam optimizer to set the learning rate to 0.001, the batch size to 64, and performs 300 rounds of iterative training on the training set. Evaluate the model performance on the validation set every 50 rounds. When the loss function of the validation set no longer decreases, apply the early stopping strategy, and then evaluate the model generalization performance on the test set to ensure that the average prediction error is less than 5%. Then use transfer learning technology to fine-tune for the air-supported structure characteristics. During the fine-tuning process, freeze the parameters of the material physical property encoder layer and only update the parameters of the acoustic response predictor and the spatial configuration optimizer. After 100 rounds of fine-tuning training, obtain the model optimized for the air-supported structure.

[0124] To better understand and implement the present invention, the following provides Example 2 of a specific application scenario of the present invention: Researchers developed a new type of light-transmitting sound-absorbing module for the acoustic renovation project of a large air-supported membrane structure stadium and successfully applied it. The air-supported membrane stadium has a building area of 12,000 square meters, and the main stadium is designed to accommodate 5,000 spectators. After completion, obvious acoustic problems such as excessive reverberation time and echo interference were found, which affected the competition and performance effects. After measurement, the reverberation time in the 500 Hz frequency band under the empty-field state reached 3.2 seconds, which met the 1.3-second limit specified in the national standard "Code for Acoustic Design and Measurement of Stadiums and Gymnasiums" (JGJ / T 131-2012). Due to the special nature of the air-supported membrane structure, conventional sound-absorbing materials cannot meet the light-transmitting requirements and have problems such as insufficient fire resistance and short service life.

[0125] The researchers adopted the method proposed by the present invention. First, non-woven polyester fiber with a light transmittance of 92% was selected as the base material, and a uniform fiber layer with a thickness of 45 mm was formed through the hot air nonwoven process, controlling the density to be 55 kg per cubic meter. The fiber raw materials were fully dispersed using the carding and opening technology, and the hot air temperature was controlled at 135 degrees Celsius to form a three-dimensional network structure with a porosity of 78%.

[0126] The fiber layer was subjected to fire retardant treatment. An aqueous solution with a mass ratio of sodium tetraborate to boric acid of 3:2 was prepared, with a concentration of 12% and a pH value of 9.0. The fiber layer was immersed in the flame retardant liquid for 8 minutes, and it was flipped every 1 minute during this period to ensure full penetration. Subsequently, it was dried at 75 degrees Celsius and a relative humidity of 35% until the moisture content reached 1.5%. After testing, the treated material reached the fire resistance B1 level standard, meeting the fire safety requirements of air-supported membrane buildings.

[0127] The treated fiber layer was prepared into a plate-shaped sound absorber with a special thickness distribution using the hot pressing forming process. The thickness of the central area of the customized mold was 30 mm, and the thickness of the edge area was 20 mm, forming a sound wave guiding type thickness gradient. The pressing temperature was set at 180 degrees Celsius, the pressure was 0.8 MPa, and the pressing time was 12 minutes. The segmented pressure increase method was adopted: initially, it was increased to 0.3 MPa at a rate of 0.1 MPa per minute and maintained for 3 minutes, and then it was increased to 0.8 MPa at a rate of 0.2 MPa per minute and maintained until the forming was completed.

[0128] The MicroPore optimization algorithm was used to calculate the micropore distribution parameters, and the acoustic characteristics and environmental parameters of the air-supported membrane structure were input, including: air density 1.20 kg / m 3 , sound speed 343 m / s, temperature 22 degrees Celsius, humidity 45%. The target sound absorption coefficients were 0.45 at 250 Hz, 0.75 at 500 Hz, 0.85 at 1000 Hz, 0.80 at 2000 Hz, 0.75 at 4000 Hz, and the required light transmittance value was 85%. The calculation results are shown in Table 1:

[0129] Table 1 Calculation Results of MicroPore Optimization Algorithm

[0130] Parameter Value Micropore diameter 0.85 mm Micropore distribution density <![CDATA[220 pieces / cm 2 > Micropore shape factor 1.2 (elliptical) Open cell ratio 32％ Predicted light transmittance 87.5％ Predicted noise reduction coefficient (NRC) 0.72

[0131] According to the calculation results, a transparent polyester microporous film with a thickness of 0.08 mm is selected, and the microporous structure is processed according to the design parameters by precision laser drilling technology and covered on the surface of the plate-shaped sound absorber. The two are compounded by ultrasonic welding process, with a welding frequency of 32 kHz, a power of 650 W, a pressure of 0.3 MPa, and a welding time of 1.2 s.

[0132] The composite material is treated for waterproofing and mildew prevention. A mixed solution of perfluorotridecyltriethoxysilane and alkyl quaternary ammonium salt with a mass ratio of 5:1 is prepared, and 3% of glycidyl methacrylate is added to adjust the pH value to 7.0, and the solid content of the solution is 10%. The composite material is immersed in the solution for 45 minutes and flipped every 10 minutes. The water droplet contact angle of the treated material reaches 135 degrees, showing excellent superhydrophobic and mildew-proof properties.

[0133] The composite material is dried by a staged drying process: the first stage is at 60 °C and a relative humidity of 40% for 40 minutes; the second stage is at 65 °C and a relative humidity of 35% for 75 minutes; the third stage is at 70 °C and a relative humidity of 30% for 35 minutes to ensure that the moisture content of the material is less than 0.8%. During the drying process, the material is laid flat on a porous support frame and flipped every 30 minutes to prevent deformation.

[0134] The acoustic performance of the composite material is analyzed using the pre-trained AcousticDome neural network model, the sound absorption coefficients in each frequency band are predicted, and the cutting plan is optimized. The comparison between the measured sound absorption coefficients and the predicted values is shown in Table 2:

[0135] Table 2 Comparison between Predicted and Measured Sound Absorption Coefficients of Composite Material

[0136] Frequency (Hz) Predicted sound absorption coefficient Measured sound absorption coefficient Relative error (%) 125 0.28 0.26 7.69 250 0.47 0.45 4.44 500 0.78 0.74 5.41 1000 0.86 0.82 4.88 2000 0.79 0.77 2.60 4000 0.76 0.73 4.11 NRC 0.73 0.70 4.29

[0137] According to the spatial geometric characteristics of the air film stadium, the composite material is cut into standard modules of 600 mm × 1200 mm, and the edges are sealed by a heat-sealing process at a temperature of 190 °C, a pressure of 0.4 MPa, and a heat-sealing time of 5 s to form a sealed edge with a width of 8 mm, and a thickness gradient structure is formed in the edge area with a reduction rate of 20%.

[0138] The completed light-transmitting sound-absorbing module was tested through an acoustic impedance tube, and the final sound-absorbing performance reached: NRC was 0.70, the average sound-absorbing coefficient in the frequency band of 500 Hz to 1000 Hz was 0.78, and the light transmittance was 87.5%, meeting the design requirements. According to the structural characteristics of the air-supported stadium, an optimal layout plan was designed. A total of 3200 square meters of light-transmitting sound-absorbing modules were installed, with a coverage rate of approximately 65% of the air-supported area above the spectator stands.

[0139] The Odeon acoustic field analysis software was used to simulate and calculate the acoustic field characteristics before and after the transformation. Based on the actual sound source and receiver position information, the number of rays was set to 15000, the reflection order was set to 2500, the scattering coefficient was set to 0.15, and the measured sound-absorbing coefficient data was used. The reverberation time was calculated using the Sabine formula: T 60 = 0.161V / A, where V is the space volume and A is the equivalent sound-absorbing area. The simulation results are shown in Table 3:

[0140] Table 3 Comparison of Acoustic Parameters of the Air-Supported Stadium before and after Transformation

[0141] Parameter Frequency (Hz) Before renovation After renovation Improvement rate (%) <![CDATA[Reverberation time T 60 (s)]]> 125 3.85 2.26 41.30 250 3.52 1.85 47.44 500 3.20 1.24 61.25 1000 3.08 1.17 62.01 2000 2.75 1.20 56.36 4000 2.25 1.05 53.33 Speech transmission index (STI) - 0.43 0.68 58.14 Sound field uniformity (dB) - 8.5 4.2 50.59

[0142] The renovated air-supported stadium was actually measured and verified during official events. The average reverberation time in the frequency band of 500 Hz - 1000 Hz in the empty-field state was 1.21 seconds, lower than the 1.3-second limit specified in the national standard "Code for Acoustic Design and Measurement of Stadiums and Gymnasiums"

[0143] (JGJ / T 131-2012). In the full-field state, the speech intelligibility index (STI) reached 0.72, reaching a "good" rating. After the installation of the light-transmitting sound-absorbing module, the indoor natural lighting conditions remained basically unchanged, and no additional lighting was required for daytime activities, saving energy consumption.

[0144] One year after the installation of the light-transmitting sound-absorbing module, a re-inspection was carried out. The performance showed no obvious attenuation, the waterproof and mildew-proof effects were good, and no mildew growth or material deformation was found, proving its applicability and durability in the air-supported structure environment. In addition, the product does not use formaldehyde-based adhesives throughout the process, and its environmental performance reaches the E0 level standard, meeting the environmental protection requirements of public buildings.

[0145] Traditional acoustic treatment methods for air-supported membrane structures mainly use suspended sound absorbers, mechanically supported sound-absorbing panels, or coated light-shielding and sound-absorbing coatings, etc., which have problems such as low light transmittance, complex installation, and short lifespan. The light-transmitting sound-absorbing module adopted in this embodiment solves the key technical problems of acoustic renovation of air-supported membrane structures and has the following remarkable improvements: First, the MicroPore optimization algorithm is used to accurately calculate the micropore parameters, achieving a balance between high light transmittance (>85%) and high sound absorption performance (NRC>0.7), and increasing the light transmittance by about 30% compared with traditional methods. Second, the ultrasonic welding process avoids the use of adhesives, maintains the integrity of the pore structure of the material, and improves the sound absorption performance by about 15%. Third, the AcousticDome neural network model predicts that the error of the acoustic performance of the material is expected to be controlled within 5%, greatly improving the design efficiency. Finally, the composite waterproof and mildew-proof treatment significantly extends the service life of the product, which is expected to reach 8-10 years, 2-3 times that of traditional products. These technological improvements enable air-supported membrane structures to meet the requirements of good daylighting and acoustic performance at the same time, providing a new solution for the acoustic renovation of public buildings such as air-supported membrane stadiums and exhibition centers.

[0146] It should be noted that the detailed explanations of the variables involved in the present invention are shown in Table 4 below.

[0147] Table 4 Variable Explanation Table

[0148]

[0149]

[0150] As described above, it is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the present invention.

Claims

1. A method for producing a light-transmitting sound-absorbing module suitable for an air film structure, characterized in that: include: Prepare a light-transmitting sound-absorbing substrate, select non-woven polyester fibers or glass fibers with high light transmittance, and form a uniform fiber layer through a hot air web-forming process; The uniform fiber layer is subjected to fireproof treatment, and the flame retardant is prepared into a flame retardant liquid for soaking treatment and drying; the treated uniform fiber layer is placed in a mold, and a plate-shaped sound absorber with a preset thickness distribution is formed by a hot pressing molding process; the micropore distribution parameters are calculated by the MicroPore optimization algorithm to output the optimal micropore diameter parameters, micropore distribution density parameters, micropore shape coefficient parameters and opening rate parameters, and a transparent microporous polymer film is covered on the surface of the plate-shaped sound absorber based on the output parameters, and the transparent microporous polymer film is composited with the plate-shaped sound absorber by an ultrasonic welding process; the composite material is prepared and treated with a waterproof and mildew-proof solution; Dry composite materials; use the pre-trained AcousticDome neural network model to analyze the acoustic performance of composite materials and optimize cutting plans; test the sound absorption coefficient of standardized modules and screen qualified products.

2. The method for producing a light-transmitting sound-absorbing module suitable for an air film structure according to claim 1, characterized in that: In the step of preparing a light-transmitting sound-absorbing substrate, the selected non-woven polyester fiber or glass fiber has a light transmittance greater than 85%, and a uniform fiber layer with a thickness of 30 to 50 mm is formed through a hot air web-forming process, and the density of the uniform fiber layer is controlled at 40 to 60 kilograms per cubic meter.

3. The method for producing a light-transmitting sound-absorbing module suitable for an air film structure according to claim 2, characterized in that: The hot air-laid process is to fully loosen the non-woven polyester fiber or glass fiber through an opening machine and then evenly spread it on a conveyor belt. Hot air with a temperature of 120 to 150 degrees Celsius is blown from the top and bottom directions at the same time to form a three-dimensional structure, thereby enhancing the overall strength of the uniform fiber layer while maintaining a high porosity to improve the sound absorption performance.

4. The method for producing a light-transmitting sound-absorbing module suitable for an air film structure according to claim 3, characterized in that: In the step of fireproofing the uniform fiber layer, a borate flame retardant is dissolved in deionized water to prepare a flame retardant liquid with a mass fraction of 10% to 15%, the uniform fiber layer is immersed in the flame retardant liquid for 5 to 10 minutes, and then dried at 70 to 80 degrees Celsius until the moisture content is less than 2%.

5. The method for producing a light-transmitting sound-absorbing module suitable for an air film structure according to claim 4, characterized in that: In the step of the hot pressing forming process, the plate-shaped sound absorber with a preset thickness distribution is formed by pressing at 170 to 190 degrees Celsius and a pressure of 0.5 to 1.0 MPa for 10 to 15 minutes.

6. The method for producing a light-transmitting and sound-absorbing module suitable for an air film structure according to claim 5, characterized in that: The transparent microporous polymer film has a thickness of 0.05 to 0.1 mm.

7. The method for producing a light-transmitting sound-absorbing module suitable for an air film structure according to claim 6, characterized in that: The ultrasonic welding process specifically uses high-frequency vibration energy to make the material molecules generate friction heat under pressure, prompting the transparent microporous polymer film to form a firm connection with the surface fibers of the plate-shaped sound absorber at the molecular level. No adhesive is used to avoid clogging the material pores and affecting the sound absorption performance. The ultrasonic frequency is 20 to 40 kHz.

8. The method for producing a light-transmitting sound-absorbing module suitable for an air film structure according to claim 7, characterized in that: In the step of preparing the waterproof and mildew-proof solution, a fluorocarbon water repellent and a quaternary ammonium salt mildew-proof agent are mixed in a mass ratio of 5 to 1, an acrylic cross-linking agent is added to adjust the pH value to 6.5 to 7.5, and the composite material of the transparent microporous polymer film and the plate-shaped sound absorber is immersed in the waterproof and mildew-proof solution for 30 to 60 minutes.

9. The method for producing a light-transmitting sound-absorbing module suitable for an air film structure according to claim 8, characterized in that: In the step of drying the composite material, the composite material is dried at 60 to 70 degrees Celsius and 30% to 40% relative humidity for 120 to 180 minutes to ensure that the composite material is dried evenly and does not deform.

10. The method for producing a light-transmitting sound-absorbing module suitable for an air film structure according to claim 9, characterized in that: In the step of applying the pre-trained AcousticDome neural network model to analyze the acoustic properties of composite materials, the sound absorption coefficient of each frequency band is predicted according to the thickness distribution, density gradient parameters, optimal micropore diameter parameters, micropore distribution density parameters, micropore shape coefficient parameters and opening rate parameters of the composite materials; the specific structure of the AcousticDome neural network model is a hybrid architecture combining a multi-layer convolutional neural network and a graph attention network, which includes three main modules: a material physical property encoder, an acoustic response predictor and a spatial configuration optimizer, wherein the material physical property encoder uses a three-layer three-dimensional convolutional neural network to extract the structural characteristics of the composite material, the acoustic response predictor uses a four-layer self-attention mechanism to establish the interaction between sound waves in different frequency bands and composite materials, and the spatial configuration optimizer considers the spatial geometric characteristics of the air membrane structure and the standardized module layout scheme through the graph neural network.

Images