A super large space sound absorbing sunroof

Abstract

The application provides a super-large space sound absorption sunroof, which comprises a window frame, a first outer glass, a middle glass and a second outer glass, the frame is provided with mounting grooves matched with the first outer glass, the middle glass and the second outer glass, the first outer glass is used for being mounted in the interior of a super-large space building, the second outer glass is used for being mounted in the exterior of the super-large space building, the thickness of the first outer glass is greater than that of the second outer glass, the first outer glass, the middle glass and the second outer glass are all made of organic glass, a large number of perforations for sound absorption are formed in the first outer glass, the distance between the first outer glass and the middle glass is 2-3 cm, and the distance between the middle glass and the second outer glass is greater than the thickness of the middle glass and the second outer glass. The technical problem that the single-layer or double-layer ordinary sound absorption window structure in the prior art cannot well realize the noise sound absorption in the super-large space is solved.

Description

Technical Field

[0001] This invention belongs to the field of sound-absorbing skylight technology, and more specifically, relates to an ultra-large space sound-absorbing skylight. Background Technology

[0002] Large buildings, industrial plants, airport terminals, subway halls, and other ultra-large spaces often suffer from severe noise pollution and acoustic vibration problems due to their massive size and complex internal structures. Noise not only affects work efficiency and production quality but also adversely impacts the physical and mental health of operators. Traditional sound insulation and noise reduction methods, such as soundproof windows and sound-absorbing felt, have limited effectiveness in ultra-large spaces and are expensive and inconvenient to maintain. Therefore, developing a new, efficient, and economical noise control solution for ultra-large spaces has significant theoretical and practical value.

[0003] Current large-space sound-absorbing skylights often use single or double-layered ordinary sound-absorbing window structures, which cannot effectively solve the technical problem of noise absorption in large spaces. Summary of the Invention

[0004] In view of this, the present invention provides an ultra-large space sound-absorbing skylight, which can solve the technical problem that current ultra-large space sound-absorbing skylights often directly use single or double-layer ordinary sound-absorbing window structures, which cannot effectively achieve noise absorption in ultra-large spaces.

[0005] This invention is implemented as follows:

[0006] This invention provides a large-space sound-absorbing skylight for use in large-space buildings, comprising a window frame, a first outer glass layer, a middle glass layer, and a second outer glass layer. The frame is provided with mounting grooves that match the first outer glass layer, the middle glass layer, and the second outer glass layer. The first outer glass layer is used for installation inside the large-space building, and the second outer glass layer is used for installation outside the large-space building. The thickness of the first outer glass layer is less than the thickness of the second outer glass layer. The first outer glass layer is made of plexiglass, and a large number of perforations for sound absorption are formed on the first outer glass layer.

[0007] Optionally, the window frame has dimensions of 300 mm × 300 mm.

[0008] Optionally, both the middle glass and the second outer glass are made of ordinary glass, and the distance between the first outer glass and the middle glass is 2-3 cm.

[0009] Optionally, the middle glass is a single layer of glass, forming a two-pane, one-cavity skylight structure with perforated organic glass for sound absorption; the thickness of the window frame is 40 mm.

[0010] Optionally, the middle glass is two layers of glass, forming a triple-glazed, two-cavity skylight structure with perforated acrylic sound-absorbing glass; the thickness of the window frame is 58 mm.

[0011] Optionally, the thickness of the middle glass layer is between the thicknesses of the first outer glass layer and the second outer glass layer.

[0012] The steps for determining the parameters of the numerous perforations used for sound absorption specifically include:

[0013] S10. Based on the Helmholtz resonance principle, determine the required sound absorption frequency range and the resonance frequency formula;

[0014] S20. Based on the preset initial perforation parameters, including hole diameter d, hole spacing D, plate thickness t, cavity depth L, and the distribution matrix of all perforations, calculate the perforation rate P and the resonant frequency f.

[0015] S30. Calculate the sound absorption coefficient at different sound absorption frequencies using a pre-trained sound absorption coefficient calculation model based on scaled-down experimental data.

[0016] S40. Calculate the noise reduction coefficient NRC based on the sound absorption coefficient at different sound absorption frequencies, and use it as an indicator to evaluate the quality of the perforation parameters.

[0017] S50. The initial perforation parameters are augmented to obtain a perforation parameter set, and then the noise reduction coefficient (NRC) of each group of perforation parameters in the perforation parameter set is obtained.

[0018] S60. Using a genetic algorithm, the initial perforation parameters are optimized to obtain multiple candidate perforation parameters with the smallest sound absorption coefficient;

[0019] S70. The optimal perforation parameter is the one corresponding to the best result of the simulation calculation of the multiple candidate perforation parameters using Zorba software.

[0020] The specific steps for determining the scaling factor in the scaling experiment include:

[0021] S31. Establish multiple sets of scaled-down simulation models of the ultra-large space sound-absorbing skylight to be tested, determine the scaling rules according to the geometric similarity theory to obtain the candidate scaling factor groups corresponding to the multiple scaled-down models, and use the genetic algorithm to optimize and obtain the optimal scaling factor group.

[0022] S32. Establish the acoustic equation of the ultra-large space sound-absorbing skylight, scale the acoustic equation using the optimal scaling factor group, obtain the acoustic parameter scaling factor of the scaled model, and add the optimal scaling factor group.

[0023] S33. Based on the optimal scaling factor group, use 3D modeling software to perform three-dimensional sound field modeling on the ultra-large space and the sound-absorbing skylight installed thereon, and obtain the three-dimensional sound field model of the scaled-down sound-absorbing skylight.

[0024] S34. Construct sound-absorbing skylight models of real size and different scales, and perform sound field numerical simulation calculations on these models.

[0025] S35. Extract acoustic characteristic parameters of sound pressure level and sound intensity level at different frequencies from the simulation results, and establish the corresponding acoustic parameter mapping relationship between the real and scaled-down models.

[0026] S36. Based on the established mapping relationship, the acoustic performance indicators of the real-size model at different frequencies, including the sound absorption coefficient, are derived from the simulation results of the scaled-down model.

[0027] S37. Compare the test data of the real model with the results of the scaled-down model to verify and correct the scaling rules, thereby obtaining the final optimized scaling factor set for building the actual scaling model.

[0028] Specifically, step S10 includes:

[0029] Step 101: Based on the Helmholtz resonance principle, obtain the resonant frequency formula of the cylindrical hole, where the formula includes integer m representing an integer multiple of the wavelength, constant γ related to the size and shape of the hole, c0 being the speed of sound in free air, ρ0 and ρ1 being the external and internal air densities respectively, and r0 being the radius of the circular hole.

[0030] Step 102: According to the formula for the resonant frequency of the cylindrical hole, the size of the hole, including the hole diameter, hole spacing, plate thickness and cavity depth, will directly affect its resonant frequency, thereby affecting the sound absorption effect.

[0031] Step 103: Based on acoustic requirements and relevant formulas, pre-determine a resonant frequency range as the design target;

[0032] Step 104: Determine the required sound absorption frequency range based on the resonant frequency range.

[0033] Specifically, step S20 includes:

[0034] Step 201: Predetermine the initial aperture, aperture spacing, plate thickness, and cavity depth;

[0035] Step 202: Obtain the distribution matrix of all holes;

[0036] Step 203: Calculate the perforation rate P based on the relationship between the aperture d and the aperture spacing D. The formula is P = πd^2 / (4D^2).

[0037] Step 204: Calculate the resonance frequency f using the Helmholtz resonance frequency formula.

[0038] Specifically, step S30 includes:

[0039] Step 301: Construct a training set by taking the measured values ​​of the actual sound absorption coefficients corresponding to different design parameters and perforation rate P obtained from the scaled-down experiment as input to the training set.

[0040] Step 302: Select a machine learning model structure, such as an artificial neural network or a support vector machine model;

[0041] Step 303: Using the training set data, the parameters of the machine learning model are trained using optimization algorithms such as backpropagation to minimize the prediction error of the model on the training set.

[0042] Step 304: Evaluate the performance of the trained model on a certain amount of test set, and adjust and optimize the model structure or training method based on the test results;

[0043] Step 305: Input the pre-set initial parameters into the trained model to obtain the corresponding predicted sound absorption coefficient values.

[0044] Specifically, step S40 includes:

[0045] Step 401: Calculate the sound absorption coefficients at four frequency points: 250Hz, 500Hz, 1000Hz, and 2000Hz.

[0046] Step 402: Calculate the noise reduction coefficient NRC using the formula NRC=(α250+α500+α1000+α2000) / 4, where α represents the sound absorption coefficient at each frequency point;

[0047] Step 403: When NRC ≥ 0.6, it is considered to have good sound absorption effect; when NRC ≥ 0.8, it is considered to have excellent sound absorption effect.

[0048] Specifically, step S50 includes:

[0049] Step 501: Use optimization algorithms such as genetic algorithms to expand the pre-set initial parameters;

[0050] Step 502: Using the pre-set initial parameters as a starting point, randomly generate a certain number of individuals within the range of each design variable to form an initial population.

[0051] Step 503: Substitute each individual into the model of step S30, calculate its NRC value, and calculate the fitness of the individual based on the NRC value.

[0052] Step 504: Select a portion of individuals from the current population according to certain rules to serve as parents for the next generation of the population;

[0053] Step 505: Cross the selected parents to generate new individuals;

[0054] Step 506: Modify some genes of the new individuals at a certain mutation rate to increase population diversity;

[0055] Step 507: Repeat steps 503 to 506 until the termination condition is met, and obtain a perforation parameter set containing multiple sets of perforation parameters.

[0056] Specifically, step S60 includes:

[0057] Step 601: Use the perforation parameter set obtained in step S50 as the initial population;

[0058] Step 602: Substitute each individual into the model of step S30, calculate its sound absorption coefficient value, and calculate the fitness of the individual based on the sound absorption coefficient value.

[0059] Step 603: Select a portion of individuals from the current population according to certain rules to serve as parents for the next generation of the population;

[0060] Step 604: Cross the selected parents to generate new individuals;

[0061] Step 605: Modify some genes of the new individuals at a certain mutation rate to increase population diversity;

[0062] Step 606: Repeat steps 602 to 605 until the termination condition is met, and obtain multiple candidate perforation parameters with the smallest sound absorption coefficient.

[0063] The specific steps for determining the scaling factor in the scaling experiment include:

[0064] Step 901: Establish multiple sets of simulated scaled-down models of the ultra-large space sound-absorbing skylight to be tested, and determine the scaling-down rules according to the geometric similarity theory to obtain the candidate scaling-down factor groups corresponding to the multiple sets of scaled-down models.

[0065] Step 902: Using optimization algorithms such as genetic algorithms, obtain the optimal scaling factor group that best describes the experimental data from the candidate scaling factor group.

[0066] Specifically, step 901 includes:

[0067] Step 1001: First, establish a 3D model at real scale, and then construct a scaled-down model according to a certain linear scaling ratio;

[0068] Step 1002: The selected scaling factor can be set to 1:1, 1:2, 1:5, 1:10, etc.

[0069] Step 1003: In addition to geometric dimensions, the scaling factor of material properties such as density and rigidity also needs to be considered.

[0070] Specifically, step S32 includes:

[0071] Step 321: Establish the acoustic equations for the ultra-large space sound-absorbing skylight;

[0072] Step 322: Scale the acoustic equation using the optimal scaling factor set to obtain the acoustic parameter scaling factor of the scaled model.

[0073] Step 323: Add the acoustic parameter scaling factor to the optimal scaling factor group.

[0074] The acoustic parameter scaling factors include sound speed scaling factors, frequency scaling factors, etc.

[0075] Specifically, step S33 includes:

[0076] Step 331: Based on the optimal scaling factor set, use 3D modeling software to perform 3D sound field modeling on the ultra-large space and its installed sound-absorbing skylights.

[0077] Step 332: Describe in detail the geometric details of key sound-absorbing structures such as sound-absorbing materials, panels, and cavities;

[0078] Step 333: Carefully model the internal structure and boundary conditions of the ultra-large space itself;

[0079] Step 334: Set appropriate sound source and receiver locations to obtain a complete scaled-down three-dimensional sound field model of the sound absorption system.

[0080] Specifically, step S34 includes:

[0081] Step 341: Construct models with different scaling scales, such as 1:1, 1:2, 1:5, and 1:10, based on the optimal scaling factor set;

[0082] Step 342: Use interpolation and other methods to accurately generate the corresponding geometric model file from the 3D model;

[0083] Step 343: Use acoustic numerical calculation software to perform sound field simulation analysis on these models to obtain acoustic response data such as sound pressure level and sound intensity level at each receiving point.

[0084] Specifically, step S35 includes:

[0085] Step 351: For each scaled model, extract acoustic parameter data such as sound pressure level and sound intensity level at different frequencies for each receiving point from the simulation calculation results.

[0086] Step 352: Normalize the frequencies of models with different scales to the same frequency coordinate system;

[0087] Step 353: Using multivariate nonlinear regression analysis, fit the nonlinear function mapping relationship between the acoustic parameters of the real model and each scaled-down model at a certain frequency point.

[0088] Specifically, step S36 includes:

[0089] Step 361: After obtaining the acoustic parameter mapping relationship between the real model and the scaled-down model at each frequency point;

[0090] Step 362: Substitute the simulation data of the scaled-down model into the mapping function to calculate the acoustic performance index value of the real model at that frequency.

[0091] Step 363: Introduce the acoustic theory model and use the acoustic performance indicators to calculate the final target quantities such as the sound absorption coefficient.

[0092] Specifically, step S37 includes:

[0093] Step 371: Manufacture a batch of full-scale sound-absorbing skylight samples and conduct standardized tests in the acoustic laboratory to obtain data such as the actual sound absorption coefficient at each frequency point;

[0094] Step 372: Compare and analyze the deviation between the measured data and the theoretical values ​​calculated based on the current scaling factor set;

[0095] Step 373: If there is a large deviation, the feedback control approach is adopted to adjust the scaling parameters according to the magnitude of the error and recalculate the new theoretical sound absorption coefficient.

[0096] Step 374: After multiple rounds of optimization calculations, an optimized scaling factor set that can accurately describe the real test data is obtained.

[0097] The scaled-down experiment includes the following apparatus: an acoustic laboratory, a loudspeaker, a microphone / sensor array, a data acquisition system, a scaled-down model, and a vibration isolation base. The acoustic laboratory is a sealed room with walls made of special sound-absorbing materials, effectively absorbing sound waves and isolating external noise. The loudspeaker emits single-frequency or swept-frequency noise signals of known frequency and amplitude as the sound source to be tested. The microphone / sensor array consists of multiple precision microphones or pressure sensors distributed at different locations to measure sound pressure level, sound intensity level, and other data at each point. The data acquisition system converts the analog signals output by the microphones into digital signals and transmits them to a computer for further processing. The scaled-down model is meticulously manufactured according to an optimized scale factor, using materials consistent with a real skylight. The vibration isolation base supports the scaled-down model in the center of the laboratory, utilizing a special vibration damping device to isolate the influence of external vibrations and clutter.

[0098] Optionally, the preparation stage of the scaled-down experiment may further include:

[0099] Step 2601: Install the scaled-down model on the vibration isolation base;

[0100] Step 2602: Arrange the sound source loudspeakers and receiving microphone array in appropriate locations according to the experimental design;

[0101] Step 2603: Check the connection status of each measuring device;

[0102] Step 2604: Calibrate the system using a standard signal source to ensure measurement accuracy.

[0103] Optionally, the scaled-down experimental measurement of the model-free baseline includes:

[0104] Step 2701: Remove all sound-absorbing materials from the laboratory;

[0105] Step 2702: Turn on the sound source speaker to emit a single-frequency or swept-frequency noise signal with a known frequency and power;

[0106] Step 2703: The microphone array collects sound pressure data at each point and records the baseline distribution of the sound field when there is no sound absorber.

[0107] Optionally, the scaled-down experimental measurement includes cases where a model is used:

[0108] Step 2801: Turn on all the basic sound-absorbing facilities in the laboratory;

[0109] Step 2802: Reposition the scaled-down sunroof model to the designated position on the vibration isolation base;

[0110] Step 2803: Transmit the same noise signal as the benchmark test;

[0111] Step 2804: The microphone array collects sound pressure data at each point again and records the sound field distribution when the model is present.

[0112] Optionally, the steps for processing the data after the scaled-down experiment include:

[0113] Step 2901: Use acoustic simulation software to establish a three-dimensional numerical model and introduce the specific structural and material parameters of the scaled-down model;

[0114] Step 2902: Import the experimental measurement data into the numerical model, and obtain refined sound field calculation results through parameter fitting;

[0115] Step 2903: Calculate the acoustic parameters of the model at different frequency points and calculate the final performance indicators such as the sound absorption coefficient.

[0116] Optionally, the steps for analyzing the scaled-down experimental results include:

[0117] Step 3001: Compare the model calculation results with the scaled-down experimental test data and analyze the magnitude of the deviation;

[0118] Step 3002: If the deviation is within the allowable range, then the set of scaling parameters can be determined as optimal;

[0119] Step 3003: If the deviation is large, the scaling rule needs to be corrected and the above experimental process repeated.

[0120] Step 3004: Based on the final determined set of optimization scaling factors, output the optimal parameters for the sunroof design.

[0121] Furthermore, the scaling factor set includes spatial size scaling factor, material density, and rigidity scaling factor.

[0122] Furthermore, the acoustic parameter scaling factors include a sound speed scaling factor and a frequency scaling factor.

[0123] The sound absorption coefficient calculation model employs an artificial neural network model, and the training steps include:

[0124] Step 1: Input the experimental data from the scaled-down experiment as the training set;

[0125] Step 2: Initialize the weights and biases of the artificial neural network;

[0126] Step 3: Perform forward propagation on the training set data and calculate the loss between the output value and the label value;

[0127] Step 4: Calculate the gradient using the backpropagation algorithm, and update the network weights and biases;

[0128] Step 5: Repeat steps 3-34 until the model converges to obtain the sound absorption coefficient calculation model.

[0129] Furthermore, the sound absorption coefficient calculation model is a support vector machine model, and the training steps include:

[0130] Input the experimental data from the scaled-down experiment as the training set;

[0131] A Gaussian kernel is used as the kernel function;

[0132] Construct and solve the dual quadratic programming problem to obtain support vectors and coefficients;

[0133] The parameters of the support vector machine model are determined using the obtained support vectors and coefficients, thus obtaining the sound absorption coefficient calculation model.

[0134] Furthermore, in the step of using a genetic algorithm to optimize and obtain the optimal scaling factor group, the initial population of the genetic algorithm is the candidate scaling factor group corresponding to the multiple scaling models.

[0135] In the step of using a genetic algorithm to optimize the initial perforation parameters, the initial population of the genetic algorithm is the initial perforation parameters.

[0136] Compared with existing technologies, the beneficial effects of the ultra-large space sound-absorbing skylight provided by this invention are:

[0137] 1. This invention establishes a two-glass-one-cavity or three-glass-two-cavity + perforated organic glass sound-absorbing skylight structure. It utilizes the cavity between the glass layers to further absorb and reduce sound. At the same time, a large number of perforations are made in the glass. When incident noise passes through these tiny holes, it will dissipate a large amount of energy through the resonance effect in the cavity, thereby achieving the purpose of sound absorption and noise reduction.

[0138] 2. Improved Prediction Accuracy of Sound Absorption Performance in Ultra-Large and Complex Sound Fields: This invention introduces a novel modeling method combining scaled-down experiments and numerical calculations. Traditional theoretical derivations and numerical simulations struggle to accurately describe the complex sound field distribution in ultra-large spaces. Scaled-down experiments, however, can reduce the model scale using geometric similarity theory, obtaining real data under controllable experimental conditions. The results under actual operating conditions can then be calculated using scaling rules. Furthermore, the scaled-down experimental data can be used to train machine learning models, further improving prediction accuracy. This invention effectively overcomes the modeling difficulties in ultra-large spaces, laying a reliable theoretical foundation for designing high-performance sound-absorbing skylights.

[0139] 3. Enhanced sound absorption performance and structural stability in harsh conditions of ultra-large spaces: This invention addresses the harsh environments of ultra-large spaces by proposing a novel cavity-filled sound-absorbing structure based on the microporous resonance principle. A special porous medium is filled between the metal substrate and the metal plate, preserving excellent sound absorption resonance effects while endowing the structure with superior heat resistance, corrosion resistance, and vibration damping properties. The porosity, material, and filling method of the porous medium have been carefully optimized to maximize the synergistic sound absorption effect. This innovative structure ensures the sound-absorbing skylight maintains stable performance under harsh conditions such as high temperature, corrosion, and vibration, making it an ideal choice for ultra-large space applications.

[0140] 4. Achieved multi-constraint global optimization of sound absorption performance: The invention employs a multi-objective optimization algorithm, using different performance indicators (such as noise reduction coefficient (NRC), sound absorption frequency range, structural strength, etc.) as optimization targets to find the globally optimal solution while satisfying various constraints. Compared with existing single-objective optimization methods, this scheme is closer to actual engineering needs, can balance different performance indicators, and achieves the best balance between sound absorption effect, structural reliability, and cost. This global optimization strategy ensures the overall excellence of the sound-absorbing skylight performance, fundamentally solving the performance defects caused by single optimization targets.

[0141] 5. Overcoming size limitations and achieving ultra-large area coverage of sound-absorbing skylights: Traditional microporous sound-absorbing structures often cannot achieve excessively large coverage areas due to manufacturing process limitations. This invention proposes an innovative modular assembly design method. Through the repeated splicing of modular units, an ultra-large area integrated sound-absorbing skylight is ultimately formed. The size of each modular unit is within the range of conventional manufacturing capabilities, but through ingenious connection methods, it can be assembled into skylight areas of any size, ensuring performance while overcoming size limitations, making it an ideal choice for solving noise control in ultra-large spaces.

[0142] In summary, the solution of this invention solves the technical problem that current large-space sound-absorbing skylights often directly use single- or double-layered ordinary sound-absorbing window structures, which cannot effectively achieve noise absorption in large spaces. Attached Figure Description

[0143] Figure 1 A schematic diagram of the two-glass, one-cavity structure of the ultra-large spatial sound-absorbing skylight provided by the present invention;

[0144] Figure 2 A schematic diagram of the triple-glass, two-cavity structure of the ultra-large spatial sound-absorbing skylight provided by the present invention;

[0145] Figure 3 A flowchart of the method provided by the present invention;

[0146] Figure 4A flowchart outlining the specific steps involved in determining the scaling factor in a scaled-down experiment.

[0147] Figure 5 This is a graph showing the sound absorption coefficient calculated using Zorba software in Example 2;

[0148] Figure 6 This is a simulation verification diagram using Odeon software for sound field analysis in Example 2;

[0149] In the attached diagram, 10 is the window frame, 20 is the first outer glass, 30 and 31 are both middle glass, and 40 is the second outer glass. Detailed Implementation

[0150] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0151] like Figure 1 The diagram shown is a schematic of an ultra-large space sound-absorbing skylight provided by the present invention. This invention provides an ultra-large space sound-absorbing skylight for ultra-large space buildings, comprising a window frame 10, a first outer glass layer 20, a middle glass layer 30, and a second outer glass layer 40. The frame is provided with mounting grooves that match the first outer glass layer, the middle glass layer, and the second outer glass layer. The first outer glass layer is used for installation inside the ultra-large space building, and the second outer glass layer is used for installation outside the ultra-large space building. The thickness of the first outer glass layer is less than the thickness of the second outer glass layer. The first outer glass layer is made of plexiglass, and numerous perforations for sound absorption are formed on the first outer glass layer.

[0152] Optionally, the window frame has dimensions of 300 mm × 300 mm.

[0153] Optionally, both the middle glass and the second outer glass are made of ordinary glass, and the distance between the first outer glass and the middle glass is 2-3 cm.

[0154] Optionally, the middle glass is a single layer of glass, forming a two-pane, one-cavity skylight structure with perforated organic glass for sound absorption; the thickness of the window frame is 40 mm.

[0155] Optional, such as Figure 2 As shown, the middle layer of glass consists of two layers of glass, 30 and 31, forming a triple-glazed, two-cavity skylight structure with perforated organic glass for sound absorption; the thickness of the window frame is 58 mm.

[0156] Optionally, the thickness of the middle glass layer is between the thicknesses of the first outer glass layer and the second outer glass layer.

[0157] The above description defines the specific structure of the ultra-large space sound-absorbing skylight provided by this invention. However, to achieve better sound absorption, simply setting perforations for sound absorption is insufficient. Currently, the widely used key technology is the "micro-pore resonant sound-absorbing structure," which utilizes the Helmholtz resonance at the entrance of the pore to absorb noise energy. The pore has a certain resonant frequency; when the incident sound wave frequency is equal to or close to the resonant frequency, pressure oscillations are generated at the pore opening, thereby dissipating most of the sound energy. The micro-pore resonant sound-absorbing structure typically consists of a layer of metal plate and a bottom cavity. The metal plate is covered with numerous tiny pores, with a pore diameter of approximately 0.1–1 mm and a pore spacing of approximately 10–50 mm. When sound waves pass through the pores, pressure oscillations are generated at the pore opening, resulting in a significant dissipation of sound energy and a reduction in sound wave transmission rate, thus achieving the purpose of sound absorption and noise reduction.

[0158] Key parameters of microporous resonant sound-absorbing structures include aperture (d), aperture spacing (D), plate thickness (t), and cavity depth (L). Different combinations of parameters will cause changes in sound absorption performance. By rationally designing these parameters, the sound absorption performance can be optimized within a specific frequency range, thereby effectively controlling certain characteristic noises. For example, to absorb single-frequency noise with a narrow frequency spectrum (such as certain mechanical rotation noise), the parameters can be optimized to have the maximum sound absorption coefficient near that frequency point; if it is necessary to absorb mixed noise with a wide frequency spectrum, the parameters should ensure that the sound absorption coefficient is uniformly distributed over a wide frequency range. Currently, the optimization of these parameters mainly relies on theoretical calculation models and finite element simulations.

[0159] In addition, the influence of cavity depth parameters needs to be considered. Deeper cavities are beneficial for lowering resonant frequencies, but they also increase manufacturing and installation costs. On the other hand, shallow cavities are not conducive to fully utilizing the sound absorption effect. Therefore, in practical engineering, the cavity depth is usually controlled within the range of 200–500 mm to balance sound absorption performance and cost requirements.

[0160] While microporous sound-absorbing structures can effectively solve noise problems in large spaces such as ordinary factories, aviation halls, and subway halls, for extra-large industrial buildings such as machine shops and laboratories, the complexity of noise, multi-source coupling, variable propagation paths, severe boundary reflections, and large variations in material properties greatly increase the difficulty of theoretical modeling and numerical calculation. Existing simulation techniques cannot accurately describe the sound absorption effect under ultra-large space conditions. Simulating the sound absorption effect in ultra-large spaces would consume too many computational resources, otherwise it would be difficult to obtain optimized perforation parameters for ultra-large space sound-absorbing skylights.

[0161] like Figure 3 The diagram shown is a flowchart of a method for determining perforation parameters used extensively for sound absorption. This method includes the following steps:

[0162] S10. Based on the Helmholtz resonance principle, determine the required sound absorption frequency range and the resonance frequency formula;

[0163] S20. Based on the preset initial perforation parameters, including hole diameter d, hole spacing D, plate thickness t, cavity depth L, and the distribution matrix of all perforations, calculate the perforation rate P and the resonant frequency f.

[0164] S30. Calculate the sound absorption coefficient at different sound absorption frequencies using a pre-trained sound absorption coefficient calculation model based on scaled-down experimental data.

[0165] S40. Calculate the noise reduction coefficient (NRC) based on the sound absorption coefficient at different sound absorption frequencies, and use it as an indicator to evaluate the quality of the perforation parameters.

[0166] S50. The initial perforation parameters are augmented to obtain a perforation parameter set, and then the noise reduction coefficient (NRC) of each group of perforation parameters in the perforation parameter set is obtained.

[0167] S60. Using a genetic algorithm, the initial perforation parameters are optimized to obtain multiple candidate perforation parameters with the smallest sound absorption coefficient;

[0168] S70. The optimal perforation parameter is the one corresponding to the best result of simulation calculation of multiple perforation parameters using Zorba software.

[0169] like Figure 4 As shown, the specific steps for determining the scaling factor in a scaling experiment include:

[0170] S31. Establish multiple sets of scaled-down simulation models of the ultra-large space sound-absorbing skylight to be tested, determine the scaling rules according to the geometric similarity theory to obtain the candidate scaling factor groups corresponding to the multiple scaled-down models, and use the genetic algorithm to optimize and obtain the optimal scaling factor group.

[0171] S32. Establish the acoustic equation of the ultra-large space sound-absorbing skylight, scale the acoustic equation using the optimal scaling factor group, obtain the acoustic parameter scaling factor of the scaled model, and add the optimal scaling factor group.

[0172] S33. Based on the optimal scaling factor group, use 3D modeling software to perform three-dimensional sound field modeling on the ultra-large space and the sound-absorbing skylight installed thereon, and obtain the three-dimensional sound field model of the scaled-down sound-absorbing skylight.

[0173] S34. Construct sound-absorbing skylight models of real size and different scales, and perform sound field numerical simulation calculations on these models.

[0174] S35. Extract acoustic characteristic parameters of sound pressure level and sound intensity level at different frequencies from the simulation results, and establish the corresponding acoustic parameter mapping relationship between the real and scaled-down models.

[0175] S36. Based on the established mapping relationship, the acoustic performance indicators of the real-size model at different frequencies, including the sound absorption coefficient, are derived from the simulation results of the scaled-down model.

[0176] S37. Compare the test data of the real model with the results of the scaled-down model to verify and correct the scaling rules, thereby obtaining the final optimized scaling factor set for building the actual scaling model.

[0177] The scaled-down experiment mainly includes the following apparatus and steps:

[0178] The device includes:

[0179] 1. Acoustics Laboratory - A sealed room with interior walls made of special sound-absorbing materials, which can effectively absorb sound waves, reduce reflection interference, and isolate the influence of external noise to ensure the purity of experimental data.

[0180] 2. Loudspeaker - capable of emitting single-frequency or swept-frequency noise signals of known frequency and amplitude as the sound source to be tested.

[0181] 3. Microphone / Sensor Array - Composed of multiple precision microphones or pressure sensors distributed at different locations to measure data such as sound pressure level and sound intensity level at each point.

[0182] 4. Data Acquisition System - Converts the analog signal output from the microphone into a digital signal and transmits it to the computer for subsequent signal processing and analysis.

[0183] 5. Scaled-down model - A meticulously crafted scaled-down model of the sunroof structure, using materials identical to those of a real sunroof, based on an optimized scale factor.

[0184] 6. Vibration Isolation Base - Supports the scaled-down model on a platform in the center of the laboratory and uses a special damping device to isolate the effects of external vibrations and clutter.

[0185] Scaled-down experimental procedure:

[0186] 1. Preparation stage

[0187] -Based on the optimized scaling factor set, a scaled-down skylight model and its spatial model are accurately produced.

[0188] - Install the scaled-down model on the vibration isolation base, and arrange the sound source loudspeakers and receiving microphone arrays in appropriate positions according to the experimental design.

[0189] - Check the connections of each measuring device and calibrate the system using a standard signal source to ensure measurement accuracy.

[0190] 2. Measurement of the baseline without a model

[0191] Remove all sound-absorbing materials (such as scale models) from the laboratory.

[0192] - Turn on the sound source speaker and emit a single-frequency or swept-frequency noise signal with a known frequency and power.

[0193] - The microphone array collects sound pressure data at each point and records the baseline distribution of the sound field when there is no sound absorber.

[0194] 3. Measurement with a model

[0195] - Turn on all the basic sound-absorbing facilities in the laboratory (such as sound-absorbing felt, vibration damping pads, etc.).

[0196] - Reposition the scaled-down sunroof model in the designated position on the vibration isolation base.

[0197] - Emits the same noise signal as the benchmark test.

[0198] - The microphone array collects sound pressure data at each point again and records the sound field distribution when the model is present.

[0199] 4. Post-processing data

[0200] -Use acoustic simulation software to build a three-dimensional numerical model and introduce the specific structural and material parameters of the scaled-down model.

[0201] -Import experimental measurement data into the numerical model and obtain refined sound field calculation results through parameter fitting.

[0202] - Calculate the acoustic parameters of the model at different frequency points and calculate the final performance indicators such as the sound absorption coefficient.

[0203] 5. Results Analysis

[0204] - Compare the model calculation results with the scaled-down experimental test data to analyze the magnitude of the deviation.

[0205] If the deviation is within the allowable range, the set of scaling parameters can be determined as optimal; otherwise, the scaling rules need to be modified and the above process repeated.

[0206] - Based on the final determined set of optimization scaling factors, output the optimal parameters for the skylight design, such as aperture d, aperture spacing D, plate thickness t, cavity depth L, etc.

[0207] In summary, scale-down experiments require a professional acoustic laboratory and supporting equipment. Sound field measurements and data analysis must be carried out according to strict procedures, and combined with numerical model simulations to ultimately determine the scale factor set that can accurately describe the real performance, providing a theoretical basis for the optimized design of ultra-large space sound-absorbing skylights.

[0208] The steps of the perforation parameter determination method in the specific implementation of the above steps are described in detail below:

[0209] The specific implementation of step S10 is as follows: based on the Helmholtz resonance principle, the resonant frequency formula for a cylindrical hole is introduced to determine the required sound absorption frequency range. Helmholtz resonance is a resonance phenomenon caused by pressure oscillations resulting from the sudden contraction or expansion of the hole's end face. Its resonant frequency can be calculated using the following formula:

[0210]

[0211] In the formula, m is an integer representing an integer multiple of the wavelength; γ is a constant related to the size and shape of the hole; c0 is the speed of sound in free air; ρ0 and ρ1 are the external and internal air densities, respectively; and r0 is the radius of the circular hole. It can be analyzed that the size of the hole (including the hole diameter d, hole spacing D, plate thickness t, and cavity depth L) directly affects its resonant frequency, thus affecting the sound absorption effect. Based on acoustic requirements and relevant formulas, a resonant frequency range can be pre-set as a design target, thereby further determining the required sound absorption frequency range.

[0212] The specific implementation of step S20 is as follows: The initial aperture d, aperture spacing D, plate thickness t, and cavity depth L, as well as the distribution matrix of all holes, are predetermined. Then, based on the perforation rate formula and resonance frequency formula for circular holes, the corresponding perforation rate P and resonance frequency f are calculated. The perforation rate P can be calculated using the following formula:

[0213]

[0214] Where d is the aperture and D is the aperture spacing. The perforation ratio P reflects the proportion of the area occupied by the pores on the sound-absorbing panel. For sound-absorbing panels with the same cavity depth, the higher the perforation ratio, the better the absorption performance of high-frequency sound waves. Therefore, the P value will be an important parameter for evaluating the design scheme. The resonant frequency f can be calculated using the Helmholtz formula mentioned earlier.

[0215] The specific implementation of step S30 is to use the sound absorption coefficient calculation model previously trained based on scaled-down experimental data to calculate the sound absorption coefficient at different frequencies for the set initial parameters. A machine learning method is used here, specifically including:

[0216] 1) Constructing a training set. Through numerous scaled-down experiments, we obtained the actual sound absorption coefficient measurements corresponding to different design parameters (d, D, t, L) and perforation rate P, and used these data as the input to the training set.

[0217] 2) Selecting a model structure. Depending on the complexity of the problem, models such as artificial neural networks and support vector machines can be used to fully exploit the nonlinear characteristics of the data.

[0218] 3) Model training. Using the training set data, optimization algorithms such as backpropagation are employed to train the model parameters, minimizing the prediction error of the model on the training set.

[0219] 4) Test and optimize. Evaluate the performance of the trained model on a certain number of test sets, and adjust and optimize the model structure or training method based on the test results.

[0220] 5) Output predicted values. Finally, input the pre-set initial parameters into the trained model to obtain the corresponding predicted sound absorption coefficient values.

[0221] The specific implementation of step S40 is to calculate the noise reduction coefficient (NRC) based on the sound absorption coefficient at different frequencies, using it as an indicator to evaluate the quality of the design parameters. The noise reduction coefficient (NRC) is a dimensionless coefficient obtained by taking the arithmetic mean of the measured sound absorption coefficients at four frequency points: 250Hz, 500Hz, 1000Hz, and 2000Hz. The formula is:

[0222]

[0223] In the formula, α represents the sound absorption coefficient at each frequency point. The larger the NRC value, the better the sound absorption performance of the design scheme for common noise frequencies. Generally, an NRC ≥ 0.6 is considered a good sound absorption effect, and an NRC ≥ 0.8 is considered an excellent sound absorption effect. In practical applications, the weight of the frequency points used to calculate NRC can be appropriately adjusted for specific noise frequency distributions.

[0224] The specific implementation of step S50 involves using an optimization algorithm to expand the pre-set initial parameters to obtain a perforation parameter set. Commonly used optimization algorithms include genetic algorithms, particle swarm optimization algorithms, and ant colony optimization algorithms. Taking the genetic algorithm as an example, the specific steps are as follows:

[0225] 1) Initialize the population. Starting with the pre-set initial parameters, a certain number of individuals are randomly generated within the range of each design variable to form the initial population.

[0226] 2) Calculate fitness. Substitute each individual into the S30 model to calculate its NRC value, and calculate the individual's fitness based on the NRC value.

[0227] 3) Selection operation. Select a subset of individuals from the current population according to certain rules to serve as parents for the next generation.

[0228] 4) Crossover operation. Crossover is performed on the selected parents to produce new individuals.

[0229] 5) Mutation manipulation. Changing some genes of new individuals at a certain mutation rate to increase population diversity.

[0230] 6) Repeat steps 2 to 5 until the termination condition is met (such as the maximum number of iterations or the fitness threshold).

[0231] Through continuous evolution using a genetic algorithm, a candidate set containing multiple sets of puncture parameters can be obtained.

[0232] The specific implementation of step S60 involves using a genetic algorithm to optimize the previously obtained set of perforation parameters, resulting in several optimized schemes with lower sound absorption coefficients. The specific steps are similar to S50, except that the objective function changes from the NRC value to the sound absorption coefficient value obtained in S30. Alternatively, other optimization algorithms such as particle swarm optimization can be used to implement this optimization process. The selection, crossover, and mutation operations in the genetic algorithm can be adjusted according to specific needs. After the optimization process terminates, several schemes with superior performance are obtained.

[0233] The specific implementation of step S70 involves using the professional acoustic modeling and simulation software Zorba to simulate and calculate the multiple previously selected optimization schemes, and selecting the parameters corresponding to the optimal result as the final design scheme. Zorba software can construct an analysis model based on the principle of Statistical Energy Analysis (SEA), considering many factors such as sound sources, sound absorption structures, and sound transmission paths, thereby calculating the indoor sound field distribution with high precision and predicting the sound absorption coefficient more accurately. After comparing and verifying the simulation results with measured data, the optimal design scheme can be output.

[0234] The specific steps for determining the scaling factor in a scaling experiment include:

[0235] The specific implementation of step S31 is to establish multiple sets of simulated scaled-down models of the ultra-large space sound-absorbing skylight to be tested, and to obtain the candidate scaled-down factor group corresponding to the multiple sets of scaled-down models by determining the scaled-down rules according to the geometric similarity theory.

[0236] The core of geometric similarity theory is that similar objects with the same linear scale ratio have identical shapes. Therefore, in designing scaled-down models, it is necessary to ensure that they have a geometrically similar shape to the actual skylight. A common approach is to first establish a 3D model at the real scale, and then construct a scaled-down model according to a certain linear scaling ratio. The possible scaling factors can be set to 1:1 (real scale), 1:2, 1:5, 1:10, etc. In addition to the geometric scale, scaling factors for material properties such as density and stiffness also need to be considered. Optimization algorithms, such as genetic algorithms, are used to enumerate different combinations of possible scaling factors, ultimately obtaining the optimal set of scaling factors that best fits the experimental data.

[0237] The specific implementation of step S32 is to establish the acoustic equation of the ultra-large space sound-absorbing skylight, scale the acoustic equation using the optimal scaling factor group to obtain the acoustic parameter scaling factor of the scaled model, and add these scaling factors to the optimal scaling factor group.

[0238] Acoustic problems can usually be described using the sound wave equation, as shown below:

[0239]

[0240] Where p is sound pressure and c is sound velocity. Based on the scale scaling coefficient and property parameter scaling coefficient in the optimal scaling factor set, the equation can be scaled accordingly to obtain the acoustic wave equation form of the scaled-down model. Furthermore, the scaling rules for key acoustic parameters such as sound velocity scaling factor and frequency scaling factor can be derived. For example, if the scale scaling coefficient is 1:n, then the frequency scaling factor is n, meaning the operating frequency of the scaled-down model will be n times that of the real model. Adding these scaling factors to the optimal scaling factor set allows for the precise application of the scaling principle in subsequent experimental measurements and simulation calculations.

[0241] The specific implementation of step S33 is as follows: based on the optimal scaling factor group, three-dimensional modeling software is used to perform three-dimensional sound field modeling on the ultra-large space and the sound-absorbing skylight installed thereon, so as to obtain a scaled-down three-dimensional sound field model of the sound-absorbing skylight.

[0242] Using professional 3D computer-aided design (CAD) software, a detailed 3D solid model of the scaled-down sound-absorbing skylight was constructed based on the scale scaling coefficients included in the optimal scaling factor set. This required meticulous description of the geometric details of key sound-absorbing structures such as sound-absorbing materials, panels, and cavities. Simultaneously, the internal structure and boundary conditions of the ultra-large space itself also needed careful modeling. Then, the scaled-down sound-absorbing skylight model was embedded into the ultra-large space model, and appropriate sound source and receiver positions were set to form a complete 3D sound field model of the scaled-down sound absorption system. This model will serve as the basis for subsequent numerical simulation calculations of the sound field.

[0243] The specific implementation of step S34 is to construct sound-absorbing skylight models of real size and different scales, and to perform sound field numerical simulation calculations on these models.

[0244] After obtaining the scaled-down 3D sound field model, models at different scales need to be constructed based on the optimal scaling factor set. Scaling scales can be selected according to ratios such as 1:1, 1:2, 1:5, and 1:10 to cover the full scale range. For each scaling scale, the corresponding geometric model file must be accurately generated from the 3D model using interpolation or other methods to serve as input for simulation calculations.

[0245] Next, acoustic numerical calculation software (such as finite element analysis software) is used to perform acoustic field simulation analysis on these models. The simulation calculation requires the introduction of corresponding acoustic governing equations, boundary conditions, etc., and the setting of a reasonable calculation grid.

[0246] Parameters such as network precision and attenuation conditions are adjusted to ensure the accuracy and convergence of the calculation results. The goal of the calculation is to obtain key acoustic response data such as sound pressure level and sound intensity level at various receiving points for different models.

[0247] The specific implementation of step S35 is to extract acoustic characteristic parameters such as sound pressure level and sound intensity level at different frequencies from the simulation results, and establish the corresponding acoustic parameter mapping relationship between the real and scaled-down models.

[0248] For each scaled model, acoustic parameters such as sound pressure level and sound intensity level at different frequencies at each receiving point can be extracted from the aforementioned simulation results. Since the frequencies corresponding to different scaled models are different (according to the scaling frequency rules), they need to be normalized to the same frequency coordinate system to establish a mapping relationship.

[0249] A commonly used modeling method is multivariate nonlinear regression analysis. Specifically, the acoustic parameters of the real model and various scaled-down models at a certain frequency can be used as independent variables, and the simulated value of the real model at that frequency as the dependent variable. A nonlinear function mapping relationship between them can then be fitted. For different acoustic parameters, the form of the mapping function may differ, requiring separate fitting. Using this mapping function, the acoustic performance indicators of the real model at the corresponding frequency can be calculated from the known acoustic data of the scaled-down models.

[0250] The specific implementation of step S36 is to deduce the acoustic performance indicators of the real-size model at different frequencies from the simulation results of the scaled-down model based on the established mapping relationship.

[0251] After obtaining the mapping relationship between the real and scaled-down models at each frequency point, the acoustic performance indicators of the real model at that frequency, such as sound pressure level, sound intensity level, and the final target sound absorption coefficient, can be calculated simply by substituting the simulation data of the scaled-down model into the mapping function. Because a complete frequency response mapping has been established, the acoustic performance curves of the real model within any desired frequency range can be calculated efficiently and accurately.

[0252] The functional relationship between the sound absorption coefficient and other acoustic parameters is usually quite complex, requiring the introduction of an acoustic theoretical model (such as a statistical acoustic energy analysis model) as an intermediate step. Specifically, the sound pressure level, sound intensity level, and other parameters of the real model are first calculated from the scaled-down results using a mapping function. Then, these parameters are substituted into the acoustic theoretical model to obtain the acoustic performance indicators of the real-size model at different frequencies, such as the sound absorption coefficient.

[0253] The specific implementation of step S37 is to compare the test data of the real model with the results of the scaled-down model, verify and correct the scaling rules, and thus obtain the final optimized scaling factor set for establishing the actual scaling model.

[0254] To verify the effectiveness of the previously determined scaling factor group, it is necessary to compare the acoustic performance of the actual model calculated by the scaling model with the actual test data. Specifically, a batch of sound-absorbing skylight samples are first manufactured at a real scale, and then standardized acoustic tests are conducted on them in a professional acoustic laboratory to obtain data such as the actual sound absorption coefficient at each frequency point.

[0255] Then, these test data are compared and analyzed for deviation with the theoretical values ​​calculated based on the current scaling factor set. If there is a large deviation, the scaling rule needs to be corrected and the scaling factor set adjusted. The correction method can adopt a feedback control approach, adjusting the scaling parameters according to the magnitude of the error, and then iteratively calculating the new theoretical sound absorption coefficient until it matches the test value.

[0256] After multiple rounds of optimization calculations, an optimized set of scaling factors that accurately describes the real test data can be obtained. This optimized set of scaling factors will serve as the basis for establishing the actual scaling model and will be used for subsequent scaling experimental research.

[0257] To better understand and implement this invention, an embodiment 1 is provided below: This ultra-large space sound-absorbing skylight adopts a double-glazed, single-cavity structure plus a perforated organic glass sound-absorbing curtain wall. Specifically, it includes the following structure:

[0258] Window frame: The window frame measures 300 mm x 300 mm and is 40 mm thick. The window frame has mounting grooves that match the first outer pane, the middle pane, and the second outer pane.

[0259] The first outer glass layer is made of acrylic glass and is 8 mm thick. This glass is installed inside the window frame, within the interior of the large-scale building. Numerous 5 mm diameter perforations are formed in the first outer glass layer, covering 20% ​​of its total area, for sound absorption.

[0260] Middle glass: The middle glass is made of acrylic glass and is 6 mm thick. The distance between the middle glass and the first outer glass is 10 mm, which is greater than the sum of the thicknesses of the two.

[0261] Second outer glass: The second outer glass is made of tempered glass and is 4 mm thick. It is installed on the outside of the window frame, located on the exterior of the large-space building. The thickness of the second outer glass is less than that of the first outer glass.

[0262] The skylight works as follows: When external noise enters the building, it first encounters the second outer layer of glass. This second outer layer is made of tempered glass, which has high rigidity and effectively blocks noise transmission. The remaining noise then enters the first outer layer of glass and the intermediate cavity. The first outer layer of glass has numerous sound-absorbing holes that absorb some of the noise energy. Simultaneously, the intermediate cavity provides some sound insulation. Through the combined effect of these two layers of glass and the intermediate cavity, the noise level inside the building is effectively reduced, creating a quiet and comfortable indoor environment for users.

[0263] The skylight features a simple and reliable design, making it easy to manufacture and install. Furthermore, through appropriate material selection and structural dimension design, it meets the sound absorption performance requirements of ultra-large spaces, providing excellent sound insulation for the building.

[0264] The following is a second embodiment of the present invention: the ultra-large space sound-absorbing skylight adopts a triple-glazed, single-cavity structure plus a perforated organic glass sound-absorbing curtain wall. Specifically, it includes the following structure:

[0265] Window frame: The window frame measures 300 mm x 300 mm and is 58 mm thick. The window frame has mounting grooves to accommodate triple-glazed windows.

[0266] The first outer glass layer is made of acrylic glass and is 10 mm thick. This glass is installed inside the window frame, within the interior of the large-scale building. Numerous 6 mm diameter perforations are formed in the first outer glass layer, covering 25% of its total area, for sound absorption.

[0267] Intermediate glass: The intermediate glass consists of two layers, each 6 mm thick, made of acrylic glass. The distance between the two intermediate glass layers is 12 mm, which is greater than the sum of the thicknesses of the two layers.

[0268] Second outer glass: The second outer glass is made of tempered glass and is 5 mm thick. It is installed on the outside of the window frame, on the exterior of the large-space building. The thickness of the second outer glass is less than that of the first outer glass.

[0269] The skylight works as follows: When external noise enters the building, it first encounters the second outer layer of glass. This second outer layer is made of tempered glass, which has high rigidity and effectively blocks noise transmission. The remaining noise then enters the two middle layers of glass and the first outer layer. The cavity between the two middle layers absorbs some of the noise energy. Simultaneously, the first outer layer has numerous sound-absorbing holes, further absorbing noise. Through the combined effect of these three layers of glass and the intermediate cavity, the noise level inside the building is effectively reduced, creating a quiet and comfortable indoor environment for users.

[0270] Compared to Example 1, this skylight adds an intermediate glass layer, further improving sound absorption performance. Simultaneously, through reasonable material selection and structural dimension design, it can meet the higher sound absorption performance requirements of ultra-large space buildings, providing excellent sound insulation for the building. While this solution slightly increases manufacturing and installation complexity, it remains a feasible design option.

[0271] Since the method for determining the perforation parameters involved in this invention involves a large amount of calculation, this part of the steps can be directly implemented in a computer. Below is a specific embodiment 3 of the method for determining perforation parameters as a computer program in a computer-readable storage medium or electronic device such as a computer. In this embodiment 3, the specific implementation of step S10 is as follows: based on the Helmholtz resonance principle, the resonance frequency formula for a cylindrical hole is introduced to determine the required sound absorption frequency range. The formula is:

[0272]

[0273] in:

[0274] f is the resonant frequency of the cylindrical hole;

[0275] m is an integer, representing an integer multiple of the wavelength;

[0276] γ is a constant that depends on the size and shape of the hole;

[0277] c0 is the speed of sound in free air, which is approximately 343 m / s;

[0278] ρ0 and ρ1 are the external and internal air densities, respectively. Under standard atmospheric pressure, the air density is approximately 1.225 kg / m^3.

[0279] r0 is the radius of the circular hole, in meters (m).

[0280] As can be seen from this formula, the dimensions of the holes, including the hole diameter d, hole spacing D, plate thickness t, and cavity depth L, directly affect the resonant frequency f, and thus the sound absorption effect. Based on acoustic requirements and relevant formulas, a resonant frequency range can be preset as a design target, thereby determining the required sound absorption frequency range.

[0281] The specific implementation of step S20 involves pre-determining the initial hole diameter d, hole spacing D, plate thickness t, cavity depth L, and the distribution matrix of all holes. Then, the perforation rate P is calculated using the perforation rate formula for circular holes.

[0282]

[0283] The perforation rate P reflects the proportion of the area occupied by the pores on the sound-absorbing panel. For sound-absorbing panels with the same cavity depth, the higher the perforation rate, the better the absorption performance of high-frequency sound waves. Next, the resonant frequency f is calculated using the Helmholtz resonant frequency formula.

[0284] The specific implementation of step S30 is to calculate the sound absorption coefficient at different frequencies using a machine learning model trained with scaled-down experimental data. Specifically, this includes:

[0285] Building the training set Where x i For different design parameters and perforation rate P as inputs, y i This corresponds to the measured value of the actual sound absorption coefficient.

[0286] Choose a model structure, such as an artificial neural network:

[0287] y = f θ (x)=φ(W L φ(W L-1 …φ(W1x+b1)…+b L-1 )+b L )

[0288] Among them W l ,b l Let φ be the weights and biases of the l-th layer, and φ be the activation function, such as the ReLU function φ(z) = max(0, z). Alternatively, a support vector machine (SVM) could be used.

[0289]

[0290] Where K(x) i (x) is a kernel function, such as the Gaussian kernel K(x) i ,x)=exp(-γ||x i -x|| 2 ).

[0291] Optimization algorithms such as backpropagation (BP) or sequence minimization (SMO) are used to train the model parameters θ on the training set to minimize the prediction error of the model on the training set.

[0292]

[0293] in For loss functions, such as squared loss Or logarithmic loss.

[0294] Evaluate the model performance on the test set and make necessary adjustments and optimizations.

[0295] Input the initial parameters into the trained model to obtain the corresponding predicted sound absorption coefficient.

[0296] The specific implementation of step S40 is to determine the sound absorption coefficient α at different frequencies. f Calculate the noise reduction coefficient (NRC):

[0297]

[0298] Where α f NRC represents the sound absorption coefficient at frequency f. The larger the NRC, the better the sound absorption performance of the design at common noise frequencies. Generally, NRC ≥ 0.6 indicates good sound absorption, and NRC ≥ 0.8 indicates excellent sound absorption.

[0299] The specific implementation of step S50 involves using optimization algorithms such as genetic algorithms to augment the initial parameters. Taking a genetic algorithm as an example:

[0300] Initialize the population P0 = {x1, x2, ..., x N}, where x i A set of parameters, randomly generated from the range of design variables.

[0301] Calculate the fitness(x) of each individual. i ) = NRC(x i ).

[0302] Selection operation: Select from P according to certain rules (such as roulette selection, tournament selection, etc.). t Select a subset of individuals as P t+1 The parent of.

[0303] Crossover operation: on P t+1 New individuals are produced by crossing over the parents. Common crossover methods include single-point crossover and multi-point crossover.

[0304] Mutation operation: with a certain mutation rate p m Changing some genes in new individuals increases population diversity. Common mutation methods include gene mutation and uniform variation.

[0305] Repeat steps 2-5 until the termination condition (such as the maximum number of generations or fitness threshold) is met, resulting in a candidate set containing multiple sets of parameters.

[0306] The specific implementation of step S60 is to use a genetic algorithm process similar to S50, take the parameter set obtained in S50 as the initial population, and perform optimization with the sound absorption coefficient as the objective function to obtain several optimization schemes with smaller sound absorption coefficients.

[0307] The specific implementation of step S70 involves using the professional acoustic modeling and simulation software Zorba to simulate and calculate the selected optimization scheme. Zorba, based on the principle of Statistical Energy Analysis (SEA), can calculate the indoor sound field distribution with high accuracy. If information such as the sound source and sound-absorbing structure is input into the Zorba model, the average acoustic energy of each subsystem can be obtained by solving the following basic equations:

[0308]

[0309] in:

[0310] π ij This represents the power transferred from subsystem i to subsystem j;

[0311] Let be the average acoustic energy of subsystem i;

[0312] τ ij is the coupling loss factor between subsystems i and j;

[0313] Let be the power injected into subsystem i.

[0314] By solving this set of equations, the average acoustic energy of each subsystem can be obtained. Then, acoustic performance indicators such as the sound absorption coefficient are calculated. The Zorba software incorporates SEA theory and provides corresponding solution algorithms. Users only need to input parameters such as model structure, materials, and boundary conditions to quickly obtain simulation results.

[0315] After comparing and verifying the simulation results with the measured data, the parameters of the optimal design scheme can be output, such as the aperture d, hole spacing D, plate thickness t, and cavity depth L.

[0316] The specific implementation of step S31 is to establish multiple sets of simulated scaled-down models of the ultra-large space sound-absorbing skylight to be tested, and to obtain the candidate scaled-down factor group corresponding to the multiple sets of scaled-down models by determining the scaled-down rules according to the geometric similarity theory.

[0317] The basic principle of geometric similarity theory is that similar objects with the same linear scale ratio have identical shapes. Let the actual size of the object be L, and the scaling scale be λ, then the size of the scaled model is λL. A common approach is to first establish a true-scale model, i.e., λ = 1, and then construct a scaled model according to a certain linear scaling ratio λ, such as λ = 1 / 2, 1 / 5, 1 / 10, etc.

[0318] In addition to geometric dimensions, material properties such as density ρ and stiffness E also need to be considered. Assuming the actual material density is ρ₀ and stiffness is E₀, then the density and stiffness of the scaled-down model are λ₀ and λ₀, respectively. 3ρ0 and λE0. Therefore, the candidate scaling factor set can be expressed as (λ, λE0). 3 ,λ).

[0319] By using optimization algorithms such as genetic algorithms to enumerate different combinations of candidate scaling factors, the optimal scaling factor set (λ) that best fits the experimental data can be obtained. * ,λ *3 ,λ * ).

[0320] The specific implementation of step S32 is to establish the acoustic equation of the ultra-large space sound-absorbing skylight, scale the acoustic equation using the optimal scaling factor group to obtain the acoustic parameter scaling factor of the scaled model, and add these scaling factors to the optimal scaling factor group.

[0321] Acoustic problems can usually be described using the sound wave equation:

[0322]

[0323] Where p is the sound pressure and c is the sound speed. Based on the scale reduction coefficient λ in the optimal scaling factor set... * And the scaling factor λ of physical property parameters *3 ,λ * The equation can be scaled accordingly.

[0324] First, multiply the variables p, x, and t by scaling factors p0 and λ, respectively. * ,λ * Dimensionless transformation of T0 yields:

[0325]

[0326] in This represents the speed of sound in the actual material. Further analysis leads to the following derivation:

[0327]

[0328] That is, the sound speed scaling factor is λ * Similarly, the frequency scaling factor f = f0 / λ can be derived. * .

[0329] Add these scaling factors to the optimal scaling factor set (λ) * ,λ *3 ,λ * ,λ * ,1 / λ * In this process, the scaling relationships of all acoustic parameters of the scaled-down model are obtained.

[0330] The specific implementation of step S33 is as follows: based on the optimal scaling factor group, three-dimensional modeling software is used to perform three-dimensional sound field modeling on the ultra-large space and the sound-absorbing skylight installed thereon, so as to obtain a scaled-down three-dimensional sound field model of the sound-absorbing skylight.

[0331] Using professional 3D computer-aided design (CAD) software, such as Creo and SolidWorks, the scale scaling coefficient λ in the optimal scaling factor set is determined. * A detailed 3D solid model of the scaled-down sound-absorbing skylight was constructed. This required precise description of the geometric details of key sound-absorbing structures such as the sound-absorbing materials, sound-absorbing panels, and cavities, ensuring complete similarity to the actual structure.

[0332] Simultaneously, the internal structure and boundary conditions of the ultra-large space itself must be carefully modeled. Then, the scaled-down sound-absorbing skylight model is embedded into the ultra-large space model, and appropriate sound source and receiver locations are set to form a complete three-dimensional sound field model of the scaled-down sound absorption system. This model will serve as the basic input for subsequent numerical simulation calculations of the sound field.

[0333] The specific implementation of step S34 is to construct sound-absorbing skylight models of real size and different scales, and to perform sound field numerical simulation calculations on these models.

[0334] Firstly, based on the optimal scaling factor set (λ) * ,λ *3 ,λ * ,λ * ,1 / λ * Construct models at different scaling scales, where the scaling scale can be selected as λ. * = 1, 1 / 2, 1 / 5, 1 / 10, etc. For each scale, the corresponding geometric model mesh file must be accurately generated from the 3D sound field model using interpolation and other methods, and used as input for simulation calculations.

[0335] Next, acoustic numerical calculation software (such as finite element analysis software COMSOL) is used to perform acoustic field simulation analysis on these models. During the simulation calculation, the acoustic control equations and boundary conditions obtained in step S32 need to be introduced, and reasonable parameters such as calculation grid accuracy and attenuation conditions need to be set to ensure the accuracy and convergence of the calculation results.

[0336] Specifically, finite element simulation requires solving governing equations, such as the Helmholtz equation for sound pressure p:

[0337]

[0338] Where k = ω / c is the wavenumber, a numerical solution can be obtained through techniques such as separation of variables and appropriate boundary conditions. The goal of the calculation is to obtain the sound pressure level L at each receiving point. p Sound intensity level LI Key acoustic response data are collected to prepare for establishing a mapping relationship in the subsequent step S35.

[0339] During simulation calculations, it is necessary to set reasonable mesh sizes and time / frequency steps based on parameters such as the model's geometry and material wave velocities to balance computational accuracy and efficiency. Simultaneously, artificial boundary conditions (such as perfectly matched layer PML or infinite element method IE) need to be set to simulate a boundaryless environment and avoid unnecessary wave reflection effects. For large models, fast algorithms such as spectral elements can also be used to accelerate computation.

[0340] Through the above simulation calculations, the acoustic response data of the model at each frequency point under different scales can be obtained, laying the foundation for establishing the mapping relationship between the real and scaled models in the subsequent step S35.

[0341] The specific implementation of step S35 is to extract acoustic characteristic parameters such as sound pressure level and sound intensity level at different frequencies from the simulation results, and establish the corresponding acoustic parameter mapping relationship between the real and scaled-down models.

[0342] For each scaling scale λ * The model allows us to extract the sound pressure level L at different frequencies f at each receiving point from the simulation results. p Sound intensity level L I Acoustic parameter data. Because the frequencies corresponding to models at different scales are different (according to the frequency scaling law f′=f / λ),... * Therefore, they need to be normalized to the same frequency coordinate system f in order to establish a mapping relationship.

[0343] A commonly used modeling method is multivariate nonlinear regression analysis, which can be specifically expressed as:

[0344]

[0345] in, These represent the sound pressure level and sound intensity level of the real model at frequency f, respectively. For the i-th scaled-down model The corresponding values ​​are: g1 and g2 are the nonlinear mapping functions to be determined; θ1 and θ2 are the mapping parameters. By optimizing the solution, the optimal mapping function can be obtained, and the acoustic parameter mapping relationship between the real and scaled-down models can be established.

[0346] The specific implementation of step S36 is to deduce the acoustic performance indicators of the real-size model at different frequencies, such as the sound absorption coefficient, from the simulation results of the scaled-down model based on the above mapping relationship.

[0347] The sound pressure level at a certain frequency f′ in the scaled-down model is known. Sound intensity level Substitute the parameters into the mapping relationship established in step S35:

[0348]

[0349] in These are estimated values ​​for the mapping parameters. The sound pressure level of the actual model at the corresponding frequency f′ can then be calculated. Sound intensity level Performance indicators, etc.

[0350] The relationship between the sound absorption coefficient α and the sound pressure level and sound intensity level is usually quite complex. A commonly used model is the acoustic theory model, such as the statistical acoustic energy analysis (SEA) model:

[0351]

[0352] Where, π ij n represents the power transferred from subsystem i to j. ij The coupling coefficient; Let be the average acoustic energy of subsystems i and j. The above calculations can be used to calculate... Substituting into the model, the sound absorption coefficient α(f) of the real-size model at different frequencies f is finally calculated.

[0353] The specific implementation of step S37 is to compare the test data of the real model with the results of the scaled-down model, verify and correct the scaling rules, and thus obtain the final optimized scaling factor set.

[0354] To verify the effectiveness of the previously determined scaling factor group, a batch of full-scale sound-absorbing skylight samples needs to be manufactured and subjected to standardized acoustic tests in a professional acoustic laboratory to obtain the true sound absorption coefficient α at each frequency point. real (f) and other data.

[0355] Then these measured data are compared with the theoretical values ​​calculated in step S36 based on the current scaling factor set. Perform comparisons and deviation analyses, such as:

[0356]

[0357] If there is a large deviation |e(f)|>∈ (where ∈ is the allowable error threshold), the scaling rule needs to be corrected. A feedback control approach can be used, adjusting the scaling parameter according to the magnitude of the error e(f) and recalculating the new theoretical sound absorption coefficient. Until the test value α real (f) Matching.

[0358] After multiple rounds of optimization calculations, the optimized scaling factor set (λ) that can accurately describe the real test data can finally be obtained.* ′,λ *′3 ,λ * ′,λ * ′,1 / λ * This optimized scaling factor set will serve as the basis for establishing the actual scaling model and will be used for subsequent scaling experimental studies.

[0359] Specifically, the principle of this invention is to integrate multiple innovative theories and advanced methods, combining new advancements in various technical fields such as optimized design, material innovation, and manufacturing processes, to form a completely new system solution. Its core technical principle is mainly:

[0360] 1. A High-Precision Sound Field Modeling Method Combining Scaled-Down Experiments and Machine Learning: Establishing accurate theoretical models and performing high-fidelity numerical calculations are extremely difficult for complex, ultra-large spatial sound fields. This invention proposes an innovative sound field modeling method that integrates the advantages of scaled-down experiments and machine learning techniques. Scaled-down experiments, based on geometric similarity theory, create reduced-scale models at different scaling ratios and obtain realistic acoustic data by controlling experimental conditions. The key is to clarify the relationships between various scaling factors, such as distance scaling, frequency scaling, and material parameter scaling, and to establish mapping formulas. These formulas normalize the measurement results of the scaled-down models to a real scale, thereby reconstructing the sound field distribution under actual working conditions with high precision. Inputting the scaled-down experimental data into machine learning algorithms can automatically uncover hidden nonlinear laws and construct more refined predictive models. For example, models based on deep neural networks can directly map the design parameters of sound-absorbing structures to the absorption coefficients at various frequencies without complex theoretical derivations. Through the close integration of scaled-down experiments and machine learning, this invention can significantly improve the modeling and prediction capabilities in ultra-large and complex sound field environments, providing reliable theoretical support for subsequent optimization design.

[0361] 2. A Multi-Objective Global Optimization Method for Sound Absorption Performance Based on Genetic Algorithm: Traditional sound absorption structure design methods have limitations, failing to simultaneously optimize multiple performance indicators and satisfy multiple constraints. This invention proposes an advanced multi-objective global optimization algorithm that can comprehensively consider different performance indicators and find the globally optimal parameter combination while satisfying various constraints. Based on genetic algorithm theory, the algorithm encodes design variables (such as aperture d, aperture spacing D, plate thickness t, cavity depth L, etc.) to construct an initial population. Each individual corresponds to a set of candidate parameter values. Next, noise reduction coefficient (NRC), sound absorption frequency range, and structural stiffness are used as multiple optimization objectives, quantified into multiple components of a fitness function. The total fitness of an individual is obtained by weighted summation of these components. During iterative optimization, selection, crossover, and mutation operations are performed based on the fitness values ​​to continuously generate a new generation of optimized population. Simultaneously, reasonable penalty functions are set for different constraints (such as process capability and cost control) to ensure that the final solution meets all constraints. Through multiple generations of population evolution, the algorithm automatically seeks the optimal solution in a trade-off space of multiple objective functions, ultimately providing a globally optimal solution that takes into account various performance indicators, serving as the parameter design scheme for the sound-absorbing skylight. The unique feature of this optimization method is that it breaks away from the limitations of traditional single-objective design drivers, truly integrating multi-indicator optimization into all aspects of sound-absorbing structure design. This ensures overall superior performance while meeting the multiple constraints of cost and reliability in practical engineering, making it a powerful tool for solving sound absorption and noise reduction problems in complex environments of ultra-large spaces.

[0362] To better understand the perforation parameter determination method involved in this invention, the following is an embodiment 4 providing a specific application scenario: A super-large stadium in a certain city is undergoing renovation and needs to install a super-large space sound-absorbing skylight. This super-large space sound-absorbing skylight adopts a double-glazed, single-cavity + perforated organic glass sound-absorbing curtain wall structure. Specifically, it includes the following structure: Window frame: The length and width dimensions of the window frame are 300 mm × 300 mm, and the thickness is 40 mm. The window frame is provided with mounting grooves that match the first outer glass, the middle glass, and the second outer glass. First outer glass: The first outer glass is made of organic glass and is 8 mm thick. This glass is installed on the inside of the window frame, located inside the super-large space building. A large number of perforations with a diameter of 5 mm are opened on the first outer glass, and the perforation area accounts for 20% of the total area of ​​the first outer glass, for sound absorption. Middle glass: The middle glass is made of organic glass and is 6 mm thick. The distance between the middle glass and the first outer glass is 10 mm, which is greater than the sum of the thicknesses of the two. The second outer glass layer is made of tempered glass and is 4 mm thick. It is installed on the outside of the window frame, located on the exterior of the large-space building. The thickness of the second outer glass layer is less than that of the first outer glass layer. The working principle of this skylight is as follows: When external noise travels into the building, it first encounters the second outer glass layer. Made of tempered glass, this layer has high rigidity and effectively blocks noise transmission. The remaining noise then enters the first outer glass layer and the intermediate cavity. The first outer glass layer has numerous sound-absorbing holes that absorb some of the noise energy. Simultaneously, the intermediate cavity provides some sound insulation. Through the combined effect of these two glass layers and the intermediate cavity, the noise level inside the building is effectively reduced, creating a quiet and comfortable indoor environment for users. The skylight features a simple and reliable design, making it easy to manufacture and install. Furthermore, through reasonable material selection and structural dimension design, it meets the sound absorption performance requirements of large-space buildings, providing excellent sound insulation for the building.

[0363] The steps for determining the perforation parameters in Example 4 are as follows:

[0364] S10. Based on the Helmholtz resonance principle, determine the required sound absorption frequency range and the resonance frequency formula.

[0365] The stadium primarily hosts sports competitions such as basketball and volleyball, as well as cultural performances. Based on noise source analysis, the main noise frequency band is concentrated in the range of 250Hz to 2kHz. To maximize the absorption of noise in this frequency band, and considering the Helmholtz resonance principle, the calculated resonant frequency range should be between 200Hz and 2.5kHz.

[0366] The specific calculation formula is as follows:

[0367]

[0368] Where m is an integer representing an integer multiple of the wavelength, ranging from 1 to 5; γ is a constant, taken as 0.3; c0 is the speed of sound in air, approximately 340 m / s; ρ0 and ρ1 are the external and internal air densities, respectively, which can be taken as 1.21 kg / m3 and 1.18 kg / m3; r0 is the radius of the circular hole, initially set to 5 mm to 20 mm.

[0369] Substituting into the above formula, we can see that when the aperture r0 is in the range of 5mm to 20mm, the required resonant frequency f is between 200Hz and 2.5kHz, which can effectively cover the target noise frequency band. Therefore, the design target sound absorption frequency range for the stadium's sound-absorbing skylight is determined to be 200Hz to 2.5kHz.

[0370] S20. Based on the preset initial perforation parameters, including hole diameter d, hole spacing D, plate thickness t, cavity depth L, and the distribution matrix of all perforations, calculate the perforation rate P and resonant frequency f.

[0371] Based on prior experience and references, the preliminary perforation parameters for the sound-absorbing skylight are determined as follows: aperture d = 10mm, hole spacing D = 50mm, plate thickness t = 20mm, and cavity depth L = 100mm. The skylight dimensions are 30m × 50m, using an equilateral triangular distribution matrix, with 600 holes per row.

[0372] Substitute into the perforation rate formula:

[0373]

[0374] The perforation rate of the initial scheme can be calculated to be P = 0.157 or 15.7%.

[0375] Substituting the parameters above into the aforementioned Helmholtz resonance frequency formula, we can obtain:

[0376]

[0377] Where m ranges from 1 to 5, the corresponding resonant frequencies are approximately 690Hz, 1380Hz, 2070Hz, 2760Hz, and 3450Hz, respectively. It can be seen that the resonant frequencies corresponding to these initial perforation parameters fall precisely within the design target frequency band, meeting the requirements.

[0378] S30. Calculate the sound absorption coefficient at different absorption frequencies using a pre-trained sound absorption coefficient calculation model based on scaled-down experimental data.

[0379] To establish a model for calculating the sound absorption coefficient, researchers first accumulated a large amount of scaled-down experimental data. The specific steps are as follows:

[0380] 1) Constructing the training set. Sound-absorbing skylight models at three scales (1:2, 1:5, and 1:10) were created, covering the main geometric parameter range of the real-world size. Acoustic tests were performed on these models, and measurement data of the sound absorption coefficients corresponding to parameters such as d, D, t, L, and P were obtained and used as training samples.

[0381] 2) Model Structure Selection. Analysis revealed a strong nonlinear relationship between the sound absorption coefficient and geometric parameters; therefore, a deep learning-based artificial neural network model was chosen. The network contains three hidden layers with 128, 64, and 32 neurons respectively, using the ReLU activation function.

[0382] 3) Model Training. The scaled-down experimental data was divided into a training set (80%) and a test set (20%). The Adam optimization algorithm was used to iteratively train the neural network parameters until the prediction error on the test set was less than 5%.

[0383] 4) Output predicted values. Input the initial perforation parameters (d=10mm, D=50mm, t=20mm, L=100mm) into the trained neural network model to obtain the predicted sound absorption coefficient values ​​in each frequency band from 200Hz to 2.5kHz.

[0384] The specific data is shown in Table 1 below (taking 0.1kHz as an example):

[0385] Table 1 Basic Data Table

[0386] Frequency (Hz) Sound absorption coefficient (α) 200 0.45 300 0.52 400 0.59 500 0.63 600 0.65 700 0.67 800 0.68 900 0.69 1000 0.70

[0387] It is evident that the initial parameter scheme exhibits relatively ideal sound absorption performance within the target frequency band.

[0388] S40. Calculate the noise reduction coefficient NRC based on the sound absorption coefficient at different sound absorption frequencies, and use it as an indicator to evaluate the quality of the perforation parameters.

[0389] According to the NRC calculation formula:

[0390]

[0391] Substituting the absorption coefficient α data at the four frequency points of 250Hz, 500Hz, 1000Hz, and 2000Hz into the equation, we obtain:

[0392]

[0393] The initial design has an NRC value of 0.66, which is within the range of good sound absorption.

[0394] S50. The initial perforation parameters are expanded to obtain a perforation parameter set, and then the noise reduction coefficient (NRC) of each group of perforation parameters in the perforation parameter set is obtained.

[0395] To further optimize sound absorption performance, a genetic algorithm was used to augment the initial parameters, generating a candidate set containing multiple sets of parameters. The specific steps are as follows:

[0396] 1) Initialize the population. Based on the aforementioned initial parameters (d = 10 mm, D = 50 mm, t = 20 mm, L = 100 mm), 20 different perforation parameters are randomly generated within the range of aperture d = 5–20 mm, aperture spacing D = 40–60 mm, plate thickness t = 15–25 mm, and cavity depth L = 80–120 mm, as the initial population.

[0397] 2) Calculate the fitness. Substitute each set of parameters into the previously trained neural network model to calculate its sound absorption coefficient in the target frequency band, thereby obtaining the NRC value. Use the NRC value as the fitness function.

[0398] 3) Selection operation. The tournament selection method is used to randomly select 4 individuals, and the 2 with the highest fitness are selected as parents.

[0399] 4) Crossover operation. For the selected parent individuals, a single-point crossover method is used to generate 2 new individuals.

[0400] 5) Mutation operation. With a 10% probability, a portion of the genes of the new individual are randomly perturbed.

[0401] 6) Repeat steps 2 to 5 for a total of 50 iterations.

[0402] After 50 generations of evolution, the genetic algorithm finally generated a set of optimization schemes containing 40 different perforation parameters. The NRC values ​​corresponding to each set of parameters are shown in Table 2 below (partial data):

[0403] Table 2. NRC Table Corresponding to Scheme Parameters

[0404]

[0405]

[0406] It can be seen that, through the optimization of the genetic algorithm, several optimal schemes with NRC values ​​between 0.6 and 0.8 have emerged in the parameter set.

[0407] S60. Using a genetic algorithm, the initial perforation parameters are optimized to obtain multiple candidate perforation parameters with the smallest sound absorption coefficient.

[0408] Based on the aforementioned 40 parameter schemes, researchers further optimized them using a genetic algorithm, with minimizing the sound absorption coefficient as the objective function. The specific steps are as follows:

[0409] 1) The aforementioned 40 schemes were used as the initial population.

[0410] 2) Calculate the fitness. Substitute each set of parameters into the trained neural network model to obtain the sound absorption coefficient α in the frequency band of 200Hz to 2.5kHz. Use the average sound absorption coefficient ā as the fitness function.

[0411] 3) Selection process. A tournament selection method is used, randomly selecting 4 individuals, and choosing the 2 with the highest fitness as parents.

[0412] 4) Crossover operation. For the selected parent individuals, a single-point crossover method is used to generate 2 new individuals.

[0413] 5) Mutation operation. With a 5% probability, a portion of the genes of the new individual are randomly perturbed.

[0414] 6) Repeat steps 2 to 5 for a total of 100 iterations.

[0415] After 100 generations of evolution, the genetic algorithm finally selected the 5 optimal schemes with the smallest sound absorption coefficients, as shown in Table 3 below:

[0416] Table 3 Optimization Scheme Table

[0417]

[0418] It can be seen that the average sound absorption coefficient of these five schemes Between 0.62 and 0.66, it is superior to the aforementioned initial scheme.

[0419] S70. The optimal perforation parameter is the one corresponding to the best result of the simulation calculation of the multiple candidate perforation parameters using Zorba software.

[0420] like Figure 5 As shown, researchers used Zorba software to conduct further acoustic simulation calculations on the five optimization schemes mentioned above. Zorba software, based on Statistical Energy Analysis (SEA) theory, can more accurately simulate the sound field distribution in ultra-large spaces, thus predicting sound absorption performance more precisely.

[0421] The specific simulation steps are as follows:

[0422] 1) Establish Zorba simulation models. Based on the specific parameters of the five optimization schemes, construct corresponding three-dimensional geometric models and set boundary conditions such as material properties, sound source location, and receiver point distribution.

[0423] 2) Solve for the sound field distribution. Solve the sound wave propagation equation using Zorba simulation to obtain the sound pressure level and sound intensity level distribution at each receiving point in the 200Hz to 2.5kHz frequency band.

[0424] 3) Calculate the sound absorption coefficient. Based on the sound field distribution data, calculate the sound absorption coefficient α for each frequency band using statistical energy analysis theory.

[0425] 4) Evaluate NRC performance. Substitute the calculated sound absorption coefficient α into the NRC formula to obtain the noise reduction coefficient (NRC) value for each scheme.

[0426] The NRC performance indicators of the five optimized schemes are shown in Table 4 below after simulation calculations:

[0427] Table 4 Simulation Optimization Scheme

[0428]

[0429]

[0430] The results show that Scheme 2 has the highest NRC value, reaching 0.72, making it the optimal design. The specific parameters of this scheme are: hole diameter d = 15mm, hole spacing D = 52mm, plate thickness t = 19mm, and cavity depth L = 95mm.

[0431] Based on the foregoing analysis, the optimal design parameters for the stadium's extra-large sound-absorbing skylight are:

[0432] Aperture d = 15mm

[0433] Hole spacing D = 52mm

[0434] Plate thickness t = 19mm

[0435] Cavity depth L = 95 mm

[0436] This parameter combination not only ensures excellent sound absorption performance (NRC = 0.72) in the target frequency band of 200Hz to 2.5kHz, but also demonstrated stability in previous scaled-down experiments, meeting the actual noise control requirements of the venue. Simulation results using Odeon software were also provided for sound field analysis, as shown in the following figures. Figure 6 As shown.

[0437] To ensure the precise implementation of this plan, researchers further conducted the following work:

[0438] 1. Create a 1:2 scale model and conduct a full-scale verification test in a professional acoustic laboratory to measure the sound absorption coefficient of each frequency band. Compare the results with the Zorba simulation results to ensure the accuracy of the model parameters.

[0439] 2. By using BIM technology, the optimized design parameters of the sound-absorbing skylights are imported into the three-dimensional architectural model of the stadium, and deeply integrated with other interior design elements, providing accurate reference for subsequent construction.

[0440] 3. Compile detailed installation process guidelines, including key steps such as the cutting dimensions of the skylight panels, hole layout, and installation sequence.

[0441] 4. In response to the characteristics of this scheme, the researchers also developed corresponding quality control measures:

[0442] (1) In terms of material selection, it is required to use plates with a thickness tolerance within ±0.5mm and a density deviation of less than 5% to ensure the consistency of resonance characteristics.

[0443] (2) In the processing and manufacturing process, the hole diameter tolerance is controlled within ±0.2mm, the hole position deviation is less than ±1mm, and the surface roughness is less than 3.2μm, so as to reduce the impact of processing errors on sound absorption performance.

[0444] (3) During installation, the hole spacing deviation must be less than ±2mm, and the flatness of the plate must be less than 2mm / m to ensure stable installation quality. At the same time, strengthen the professional training of construction personnel and standardize the operating procedures.

[0445] (4) During the final acceptance, acoustic tests are conducted on the sample area to measure the sound absorption coefficient of each frequency band and compare it with the design indicators. Only after the standards are met can the sample be accepted.

[0446] Through the quality control of the entire process described above, we can ensure that this extra-large sound-absorbing skylight can reliably achieve the expected noise control effect in practical applications.

[0447] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for determining the perforation parameters of an ultra-large space sound-absorbing skylight, wherein, The super-large space sound-absorbing skylight includes a window frame, a first outer glass layer, a middle glass layer, and a second outer glass layer. The window frame is provided with mounting grooves that match the first outer glass layer, the middle glass layer, and the second outer glass layer. The first outer glass layer is used to be installed inside the super-large space building, and the second outer glass layer is used to be installed outside the super-large space building. The thickness of the first outer glass layer is less than the thickness of the second outer glass layer. The first outer glass layer is made of plexiglass and has a large number of perforations for sound absorption. The method for determining the parameters is characterized by comprising: S10. Based on the Helmholtz resonance principle, determine the required sound absorption frequency range and the resonance frequency formula; S20. Based on the preset initial perforation parameters, including hole diameter d, hole spacing D, plate thickness t, cavity depth L, and the distribution matrix of all perforations, calculate the perforation rate P and resonant frequency f; S30. Calculate the sound absorption coefficient at different sound absorption frequencies using a pre-trained sound absorption coefficient calculation model based on scaled-down experimental data. S40. Calculate the noise reduction coefficient (NRC) based on the sound absorption coefficient at different sound absorption frequencies, and use it as an indicator to evaluate the quality of the perforation parameters; S50. The initial perforation parameters are augmented to obtain a perforation parameter set, and then the noise reduction coefficient (NRC) of each group of perforation parameters in the perforation parameter set is obtained. S60. Using a genetic algorithm, the initial perforation parameters are optimized to obtain multiple candidate perforation parameters with the smallest sound absorption coefficient; S70. The optimal perforation parameter is the selected perforation parameter corresponding to the optimal result of the simulation calculation of the multiple candidate perforation parameters using Zorba software, and is output as the optimal perforation parameter. The sound absorption coefficient calculation model employs an artificial neural network model, and the training steps include: Step 1: Input the experimental data from the scaled-down experiment as the training set; Step 2: Initialize the weights and biases of the artificial neural network; Step 3: Perform forward propagation on the training set data and calculate the loss between the output value and the label value; Step 4: Calculate the gradient using the backpropagation algorithm, and update the network weights and biases; Step 5: Repeat steps 3-34 until the model converges to obtain the sound absorption coefficient calculation model; The specific steps for determining the scaling factor in the scaling experiment include: S31. Establish multiple sets of scaled-down simulation models of the ultra-large spatial sound-absorbing skylight to be tested. Determine the scaling-down rules according to the geometric similarity theory to obtain multiple sets of candidate scaling-down factor groups corresponding to the scaled-down models. Use a genetic algorithm to optimize and obtain the optimal scaling-down factor group. The scaling-down factor group includes spatial size scaling-down factor, material density and rigidity scaling-down factor. S32. Establish the acoustic equation of the ultra-large space sound-absorbing skylight, scale the acoustic equation using the optimal scaling factor group to obtain the acoustic parameter scaling factor of the scaled model, and add the optimal scaling factor group, wherein the acoustic parameter scaling factor includes the sound speed scaling factor and the frequency scaling factor. S33. Based on the optimal scaling factor group, use 3D modeling software to perform three-dimensional sound field modeling on the ultra-large space and the sound-absorbing skylight installed thereon, and obtain the three-dimensional sound field model of the scaled-down sound-absorbing skylight. S34. Construct sound-absorbing skylight models of real size and different scales, and perform sound field numerical simulation calculations on these models. S35. Extract the acoustic characteristic parameters of sound pressure level and sound intensity level at different frequencies from the simulation results, and establish the corresponding acoustic parameter mapping relationship between the real and scaled-down models. S36. Based on the established mapping relationship, the acoustic performance of the real-size model at different frequencies, including the sound absorption coefficient, is deduced from the simulation results of the scaled-down model. S37. Compare the test data of the real model with the results of the scaled-down model to verify and correct the scaling rules, thereby obtaining the final optimized scaling factor set for building the actual scaling model.

2. The method for determining the perforation parameters of the ultra-large space sound-absorbing skylight according to claim 1, characterized in that, Both the middle glass and the second outer glass are made of ordinary glass, and the distance between the first outer glass and the middle glass is 2-3 cm.

3. The method for determining the perforation parameters of the ultra-large space sound-absorbing skylight according to claim 1, characterized in that, The middle glass is a single layer of glass, forming a two-glass, one-cavity + perforated organic glass sound-absorbing skylight structure.

4. The method for determining the perforation parameters of the ultra-large space sound-absorbing skylight according to claim 1, characterized in that, The middle glass consists of two layers of glass, forming a triple-glazed, two-cavity skylight structure with perforated organic glass.

5. The method for determining the perforation parameters of the ultra-large space sound-absorbing skylight according to claim 1, characterized in that, The thickness of the middle glass layer is between the thickness of the first outer glass layer and the thickness of the second outer glass layer.

6. The method according to claim 1, characterized in that, The window frame measures 300 mm in length and 300 mm in width.

7. The method for determining the perforation parameters of an ultra-large spatial sound-absorbing skylight according to claim 3, characterized in that, The thickness of the window frame is 40 mm.

8. The method for determining the perforation parameters of the ultra-large space sound-absorbing skylight according to claim 4, characterized in that, The thickness of the window frame is 58 mm.

Images