Arrayed acoustic black hole vented sound barrier and application system

Abstract

The application relates to the technical field of acoustic black hole sound insulation structure, in particular to an array type acoustic black hole ventilation sound insulation barrier and an application system.In the application, a generative adversarial network is used to extract multiple frequency band values, and a random distribution noise sequence is used to perform multiplication and addition algebraic operation, so that precise mode matching of a nonlinear thickness evolution gradient and a wideband sound wave energy distribution is realized, the energy trap capture bandwidth of the structure in a low and medium frequency band is significantly improved, a genetic algorithm is used to extract a damping groove space coordinate characteristic segment and set a crossover and mutation probability threshold, so that the sound energy loss efficiency reaches a convergence extreme value in a complex topological node grid, the traditional point arrangement scheme is solved under multi-dimensional constraints to avoid falling into a local engineering problem, environmental characteristic vectors are constructed by splicing airflow and temperature gradient sequences, a curve loss difference value is measured and compared with a warning limit, the performance degradation trajectory of the barrier is accurately captured, and the system can maintain stable sound insulation gain under complex airflow and thermal field interference.

Description

Technical Field

[0001] This invention relates to the field of acoustic black hole sound insulation structure technology, and in particular to array-type acoustic black hole ventilation sound insulation barrier and application system. Background Technology

[0002] The core mechanism of acoustic black hole sound insulation structure technology is to use a specific geometric shape, specifically a region with decreasing thickness according to a power law, to induce a reduction in wave velocity of curved waves in the region. When the wave propagates to the tip of the structure, the wave velocity drops to zero and no reflection occurs, thereby achieving a high concentration of energy in the local area. The research focus of this technology includes multiple acoustic black hole structure damping layer laying schemes, multi-cell array coupling effects, and broadband vibration energy dissipation algorithms. This field does not rely on simple physical isolation, but uses the structural dynamics to achieve an energy trapping effect, solving the trade-off between structural lightweighting and low-to-mid-frequency sound insulation performance.

[0003] The array-type acoustic black hole ventilation sound barrier and its application system is a spatial array structure composed of multiple acoustic black hole functional units. Specifically, it is characterized by embedding a thinning structure with a thickness that decays exponentially to near zero on the surface of the ventilation barrier blades and the inner lining wall while maintaining airflow channels. The design aims to resolve the conflict between noise leakage and air resistance in open environments of ventilation equipment. The primary purpose of the system is to construct a nonlinear sound energy attenuation channel. By capturing sound waves of specific frequency bands through the acoustic black hole region, the sound energy is directed to a high-damping loss material covering layer. The solution achieves a reduction in low-to-mid frequency sound pressure level of 10 to 25 dB while ensuring that the ventilation volume reduction rate is less than 15%, and avoids regenerated noise interference generated by traditional resistive silencing devices under high-speed airflow impact.

[0004] In practical operation, existing technologies primarily rely on empirical power-law formulas to determine structural thickness. This leads to an adaptation discrepancy between the acoustic black hole tip structure and actual manufacturing processes. When airflow velocity and ambient temperature fluctuate, the matching relationship between material damping characteristics and sound wave frequency shifts. Due to a lack of dynamic perception of environmental characteristic vectors, the structure often fails to maintain the initial sound insulation performance. Conventional ventilation barriers only focus on the single dimension of the blade surface, neglecting the acoustic impedance correlation mechanism between multi-cell arrays. Sound waves easily penetrate through high-impedance gaps, significantly weakening the sound energy accumulation effect. Furthermore, the lack of quantitative screening of spatial array and volume limits often restricts structural layout due to physical space interference, resulting in insufficient effective area of ​​the acoustic black hole functional region. For example, in industrial high-temperature ventilation applications, existing technologies lack a failure time calibration mechanism. When the structure undergoes slight deformation due to thermal stress and the damping material degrades, the system cannot provide timely warnings of performance degradation. This leads to an exponential increase in noise leakage over operating time, severely interfering with the quality of the surrounding acoustic environment. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and propose an array-type acoustic black hole ventilation sound barrier and its application system.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: an array-type acoustic black hole ventilation and sound barrier and its application system include: Boundary configuration generation module: By measuring the length and width benchmarks of the outer frame and the constant of the louver assembly spacing, the module splices the values ​​to construct a spatial position array, calculates the element product to output the volume parameter and compares the parameter with the volume upper limit, removes out-of-limit combinations and summarizes the remaining items to form the boundary dimension feature set. Contour thickness extrapolation module: Based on the boundary dimension feature set, a generative adversarial network is used to extract frequency band values ​​and perform multiply-accumulate operations in conjunction with noise to output a discrete thickness sequence. After mapping the sequence to the sharpening area to calculate the frequency response matrix, the constrained error sequence is extracted and spliced ​​to form a thinning curvature array. Spatial array arrangement module: Based on the thinned curvature array, the damping groove coordinates are extracted and the distance is calculated and arranged in descending order. The first column is taken as the parent feature. The genetic algorithm is used for crossover and mutation to derive the offspring and calculate to the limit point. The extreme value coordinates are projected to the ventilation opening and the topological node coordinate system is output. Coupled sound field calculation module: Based on the thinned curvature array and the topological node coordinate system, extract the nodes and thickness, calculate the reciprocal to construct the edge weights, extract the patch loss factor and then merge the parameters to perform superposition, aggregate the features to perform dimensionality reduction mapping and extract the test point values ​​to form a sound pressure attenuation matrix; Performance degradation prediction module: Based on the sound pressure attenuation matrix, the airflow and temperature gradient sequences are extracted and merged to construct an environmental feature vector. The vector is spliced ​​with the time state parameters, the curve loss difference is calculated and compared with the warning limit. The over-limit time series is extracted to obtain the failure time calibration value.

[0007] As a further embodiment of the present invention, the boundary dimension feature set includes multiple outer frame measurement length and width references, multiple louver assembly spacing constants, and multiple compliant dimension retention items; the thinning curvature array includes multiple thickness discrete sequences, multiple constrained error sequences, and multiple compliant discrete sequence distribution node combinations; the topological node coordinate system specifically includes multiple damping groove coordinates, multiple extreme value coordinates, and multiple compliant node parameters; the sound pressure attenuation matrix includes multiple node connectivity weight sets, multiple patch loss factors, and multiple discrete frequency band test point measurement values; and the failure time calibration value specifically includes multiple environmental feature vectors, multiple curve loss differences, and multiple exceedance time series.

[0008] As a further aspect of the present invention, the boundary configuration generation module includes: Array construction submodule: Based on the outer frame measurement length and width benchmark and the louver assembly spacing constant, extract data, align and stitch to construct coordinate nodes, accumulate span values ​​to arrange spatial array, calculate the array element product output parameters, and generate spatial volume matrix; Parameter identification submodule: Based on the spatial volume matrix, read the matrix sequence to extract individual volume parameters, compare the parameter with the upper limit value of the volume, eliminate the size combination corresponding to the out-of-bounds value, reorganize the compliant size summary sequence, and generate the boundary dimension feature set.

[0009] As a further aspect of the present invention, the contour thickness estimation module includes: Dimensional Inference Submodule: Based on the boundary dimension feature set, a generative adversarial network is used to extract multiple frequency band constraint values, and a multiply-accumulate algebraic operations are performed in conjunction with a randomly distributed noise sequence to sort out and output the thickness distribution parameters of each node section, summarize and integrate the multi-dimensional thickness distribution values, and generate a discrete thickness sequence. Response calculation submodule: Based on the thickness discrete sequence, extract and map the parameters of each sequence node to the edge sharpening region space, calculate and output the normal impedance response value of the physical surface region, integrate and reorganize the multi-band calculation parameters to construct a feature distribution combination, and establish the edge frequency response matrix; Error splicing submodule: Based on the edge frequency response matrix, extract and compare the parameter distribution of the benchmark physical sample, calculate the absolute difference of each error, screen and extract the compliant discrete sequence corresponding to the limited error range, splice the distribution node combination of each compliant discrete sequence, and obtain the thinned curvature array.

[0010] As a further aspect of the present invention, the generative adversarial network extracts multiple damping groove spatial coordinate feature segments through the reference parent parameters, sets a crossover probability threshold parameter, compares the randomly generated values ​​with the crossover probability threshold parameter, swaps the coordinate data segments associated with the reference parent parameters, obtains the initial child coordinate sequence, extracts the multidimensional spatial parameters of the initial child coordinate sequence, sets a mutation probability threshold parameter, compares the randomly generated values ​​with the mutation probability threshold parameter, applies a perturbation numerical bias term to the multidimensional spatial parameters, superimposes the mutation perturbation values, and derives the child parameters.

[0011] As a further aspect of the present invention, the steps include extracting and mapping various sequence node parameters to the edge sharpening region space, extracting multiple thickness discrete sequence node parameters, setting the origin value of the physical boundary coordinates of the edge sharpening region, extracting the spatial step span value, dividing multiple sharpening three-dimensional spatial grids, extracting multiple grid position index parameters, extracting multiple thickness discrete sequence associated sequence indexes, comparing multiple associated sequence indexes with multiple grid position index parameters, matching multiple spatial grids corresponding coordinate nodes, and projecting multiple thickness discrete sequence node parameters to multiple corresponding coordinate nodes.

[0012] As a further aspect of the present invention, the spatial array arrangement module includes: Parametric sorting submodule: Based on the thinned curvature array, extract the spatial coordinate data of the damping groove, calculate the coordinate distance difference parameter, summarize the difference calculation results and sort them in descending order, extract the parent features from the top sorted coordinate data, and generate the benchmark parent parameter; Perturbation-derived sub-module: Based on the baseline parent parameters, a genetic algorithm is used to extract feature fragments and perform crossover, superimpose mutation perturbation values ​​to derive sub-parameters, iteratively calculate the sub-parameters to the limit point, extract the extreme value coordinate data corresponding to the calculation limit point, and establish an extreme value coordinate set. Coordinate mapping submodule: Based on the extreme value coordinate set, it projects the node distribution values ​​to the grid system, matches the grid entity boundary constraint parameters, compares the constraint parameters to eliminate spatially conflicting node values, summarizes and retains compliant node parameters, and outputs the topology node coordinate system.

[0013] As a further aspect of the present invention, the genetic algorithm extracts multiple spatial coordinate feature segments through the baseline parent parameters, sets a crossover probability threshold parameter, compares the size of the randomly generated values ​​with the crossover probability threshold parameter, extracts the corresponding position values ​​of the spatial coordinate feature segments, swaps the sequence parameters within the spatial coordinate feature segments, merges the swapped multiple sequence parameters, and obtains the initial offspring coordinate sequence.

[0014] As a further aspect of the present invention, the coupled sound field calculation module includes: Parametric overlay submodule: Based on the thinned curvature array and the topological node coordinate system, extract the node distribution parameter and thickness parameter, calculate the reciprocal parameter of node distance and establish cell connectivity edge weights, extract the patch loss factor, merge the factor parameter and edge weights to perform overlay algebra operation, and generate a set of node connectivity weights; Feature aggregation submodule: Based on the node connectivity weight set, it aggregates the spatial distribution feature parameters of the nodes, performs feature-level dimensionality reduction mapping operation, extracts the corresponding measured values ​​of discrete frequency band test points and summarizes the measured values ​​of the points to form a sound pressure attenuation matrix.

[0015] As a further aspect of the present invention, the performance degradation prediction module includes: Vector feature splicing submodule: Based on the sound pressure attenuation matrix, extract the airflow sequence and the windward temperature gradient sequence, merge the two parameters to construct an environmental feature vector, splice the vector with the time node parameter, and establish an evolution state feature set; Loss difference calculation submodule: Based on the evolution state feature set, calculate the absolute loss difference between the attenuation and calibration curve values, compare the loss difference with the system warning limit parameter and extract the time series data corresponding to the exceedance limit to obtain the failure time calibration value.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, a generative adversarial network is used to extract multiple frequency band values ​​and perform multiply-add algebraic operations on a randomly distributed noise sequence. The thickness discrete sequence is output and mapped to the sharpened region to calculate the frequency response matrix. This achieves accurate modal matching between the nonlinear thickness evolution gradient and the broadband acoustic energy distribution, significantly improving the energy trapping bandwidth of the structure in the low and mid-frequency bands. In this invention, a genetic algorithm is used to extract the spatial coordinate feature fragments of the damping groove and set the crossover and mutation probability thresholds. The reference parent parameter associated coordinate data are interchanged and a perturbation numerical bias term is applied to achieve a global heuristic collaborative search of the distribution law of damping material and the spatial constraint of ventilation channel. This enables the sound energy loss efficiency to reach the convergence extreme value in the complex topological node grid, solving the engineering problem that traditional point layout schemes get stuck in local conditions under multidimensional constraints. In this invention, an environmental feature vector is constructed by splicing airflow and temperature gradient sequences, the curve loss difference is calculated and compared with the warning limit, and the failure time calibration value is obtained. This enables precise capture of the barrier performance degradation trajectory and ensures that the system maintains stable sound insulation gain under complex airflow and thermal field interference. Attached Figure Description

[0017] Figure 1 This is a system flowchart of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0019] Example 1 Please see Figure 1 This invention provides a technical solution, comprising an array-type acoustic black hole ventilation and sound insulation barrier and its application system, including: Boundary configuration generation module: By measuring the length and width benchmarks of the outer frame and the constant of the louver assembly spacing, the module splices the values ​​to construct a spatial position array, calculates the element product to output the volume parameter and compares the parameter with the volume upper limit, removes out-of-limit combinations and summarizes the remaining items to form the boundary dimension feature set. Contour thickness extrapolation module: Based on the boundary dimension feature set, it uses a generative adversarial network to extract frequency band values ​​and performs multiply-accumulate operations with noise to output a discrete thickness sequence. After mapping the sequence to the sharpening area to calculate the frequency response matrix, it extracts the constrained error sequence and splices it to build a thinning curvature array. Spatial array arrangement module: Based on the thinned curvature array, the damping groove coordinates are extracted and the distance is calculated and arranged in descending order. The first column is taken as the parent feature. The genetic algorithm is used for crossover and mutation to derive the offspring and calculate to the limit point. The extreme value coordinates are projected to the ventilation opening and the topology node coordinate system is output. Coupled sound field calculation module: Based on the thinned curvature array and topological node coordinate system, extract nodes and thickness, calculate the reciprocal to construct edge weights, extract patch loss factors and merge parameters to perform superposition, aggregate features to perform dimensionality reduction mapping and extract test point values ​​to form a sound pressure attenuation matrix; Performance degradation simulation module: Based on the sound pressure attenuation matrix, the airflow and temperature gradient sequences are extracted and merged to construct an environmental feature vector. The vector is spliced ​​with the time state parameters, the curve loss difference is calculated and compared with the warning limit. The over-limit time series is extracted to obtain the failure time calibration value.

[0020] The boundary dimension feature set includes multiple outer frame measurement length and width benchmarks, multiple louver assembly spacing constants, and multiple compliant dimension retention items. The thinning curvature array includes multiple thickness discrete sequences, multiple constrained error sequences, and multiple compliant discrete sequence distribution node combinations. The topology node coordinate system specifically includes multiple damping groove coordinates, multiple extreme value coordinates, and multiple compliant node parameters. The sound pressure attenuation matrix includes multiple node connectivity weight sets, multiple patch loss factors, and multiple discrete frequency band test point measurement values. The failure time calibration values ​​specifically include multiple environmental feature vectors, multiple curve loss differences, and multiple exceedance time series.

[0021] The boundary configuration generation module includes: Array construction submodule: Based on the outer frame measurement length and width benchmark and the louver assembly spacing constant, extract data, align and stitch to construct coordinate nodes, accumulate span values ​​to arrange spatial array, calculate the array element product output parameters, and generate spatial volume matrix; Parameter identification submodule: Based on the spatial volume matrix, read the matrix sequence to extract individual volume parameters, compare the parameter with the upper limit value of the volume, eliminate the size combination corresponding to the out-of-bounds value, reorganize the compliant size summary sequence, and generate the boundary dimension feature set; The array construction submodule is based on the outer frame measurement of 2000 mm and 1500 mm in length and width, and the 50 mm spacing constant of the louver assembly. It uses three-dimensional affine transformation matrix operation to extract data, align and stitch to construct coordinate nodes, establishes a 4x4 homogeneous coordinate transformation matrix, sets the X-axis translation vector parameter to 10 mm and the Y-axis rotation angle parameter to 15 degrees, applies matrix multiplication operation to update multiple coordinate values, accumulates span values ​​to arrange the spatial array, uses a multidimensional array broadcast mechanism to configure a 200x150x30 three-dimensional tensor, assigns a data type of 64-bit floating point, executes item-by-item multiplication calculation instructions on multiple tensor elements inside the array, outputs multiple 15000 cubic millimeter single-unit volume parameters, integrates the multiple multiplication calculation result data sequence and reshapes the tensor dimension to a two-dimensional storage structure to generate a spatial volume matrix. The parameter identification submodule extracts individual volume parameters based on the spatial volume matrix by reading the multidimensional matrix slice sequence. A preset volume upper limit of 25,000 cubic millimeters is introduced as a benchmark comparison boundary. This preset volume upper limit is set by retrieving data from the physical wind tunnel boundary limit library and associating it with a product tolerance coefficient of 0.8. A quicksort algorithm is used to select the middle physical memory index position as the benchmark pivot. Bidirectional pointers are used to traverse the left and right subarrays. A step-by-step recursive comparison is performed between multiple extracted parameters and the above benchmark comparison boundary value. Boolean mask filtering is used to assign Boolean false value labels to array cells containing values ​​exceeding the 25,000 cubic millimeter limit. The memory reclamation mechanism based on the label labels is used to remove the size combinations corresponding to the out-of-bounds values. The Boolean true value labels are traversed and extracted and associated with the original array physical index position. Based on the extracted index positions, the initial matrix memory block data is read and reconstructed into a compliant size summary sequence. Multiple compliant sequences are combined and entered into a hash table data structure to generate a boundary dimension feature set.

[0022] The contour thickness extrapolation module includes: Dimensional Inference Submodule: Based on the boundary dimension feature set, it uses a generative adversarial network to extract multiple frequency band constraint values, and performs multiply-accumulate algebraic operations in conjunction with a randomly distributed noise sequence to sort out and output the thickness distribution parameters of each node section, summarize and integrate the multidimensional thickness distribution values, and generate a discrete thickness sequence. The response calculation submodule extracts and maps the parameters of each sequence node to the edge sharpening region space based on the thickness discrete sequence, calculates and outputs the normal impedance response value of the physical surface region, integrates and recombines the multi-band calculation parameters to construct the feature distribution combination, and establishes the edge frequency response matrix. Error splicing submodule: Based on the edge frequency response matrix, extract and compare the parameter distribution of the benchmark physical sample, calculate the absolute difference of each error, screen and extract the compliant discrete sequence corresponding to the limited error range, splice the distribution node combination of each compliant discrete sequence to obtain the thinned curvature array; The dimensional inference submodule, based on the boundary dimension feature set, employs a deep convolutional generative adversarial network algorithm to extract multiple frequency band constraint values. It then calls a deep learning programming framework module to construct a two-layer model topology architecture consisting of a generator and a discriminator. The generator is configured to perform sequence feature upsampling and extension calculations using a four-level inverse one-dimensional convolutional network. Each network level is assigned a physical size parameter of 5 for the convolutional kernel and a span parameter of 2. Normalization is embedded to correct the feature tensor distribution offset at the batch level. A negative half-axis control slope parameter of 0.2 is applied, and a leaky linear rectified function is used to activate the mapping space. A preset standard normal distribution random parameter tensor is extracted and imported. This preset standard normal distribution random parameter tensor refers to the underlying hardware pseudo-parameter tensor. A random floating-point generator directly generates a one-dimensional array with a mathematical expectation of 0 and a standard deviation of 1, and fixes the corresponding memory address. It uses the element-wise multiplication and addition calculation instructions of the underlying tensor matrix to jointly import the random distribution noise sequence and perform multiplication and addition algebraic operations. The result of the multiplication and addition calculation is input to the discriminator at the fully connected level to calculate the probability between 0 and 1 and determine the floating-point value. The tensor shape reshaping instruction is called to convert the one-dimensional sequence to a two-dimensional matrix shape, sort out and extract and output the thickness distribution parameters of each node section. The high-dimensional array concatenation and splicing instruction is used to merge multiple floating-point record entries along the specified dimensional axis to summarize and integrate multi-dimensional data to generate multi-dimensional thickness distribution values. The various feature vectors are assembled and directly output to generate a discrete thickness sequence. The response measurement submodule, based on the discrete thickness sequence, uses a bicubic spline interpolation algorithm to extract and map the parameters of each sequence node to the edge-sharpening region space. It calls the internal grid data interpolation mapping function module of the scientific computing core package, inputting the original one-dimensional spatial coordinate sequence nodes as a set of discrete independent variables. It constructs a two-dimensional dense solution grid space with a resolution span parameter of 0.01 mm, performs high-order polynomial surface fitting to smoothly transition the differences of discrete thickness parameters, extracts and maps the interpolation points, and establishes the topological morphology of the continuous physical space curvature distribution. It uses a fourth-order Runge-Kutta numerical integration algorithm to calculate and output the normal impedance response value of the physical surface region, setting the single-iteration time step parameter to 0. 0.005 milliseconds, initial boundary displacement parameter value of 0 mm and initial vibration velocity parameter value of 0 mm / s, through 10,000 consecutive iterations of ordinary differential equation system to solve and accumulate multi-term transient response state variable array, extract multi-order state space derivative parameters and apply weight coefficients in the range of 0.166 to 0.333 for linear superposition algebra calculation, call the complex domain impedance mapping calculation subroutine to convert the time domain state variable array to the frequency domain complex array, call the high-dimensional array stacking and merging function instruction to integrate and reorganize the multi-term band measurement corresponding complex parameter array to establish feature distribution combination, assemble the multi-channel data structure according to column priority order and directly output to construct the edge frequency response matrix; Error splicing submodule: Based on the edge frequency response matrix, it calls the underlying file input / output read instructions to extract and read a preset benchmark physical sample parameter distribution data file. This preset benchmark physical sample parameter distribution data file refers to a read-only form record document format stored in the system root directory database after the system has undergone three on-site wind tunnel acoustic acquisition platform tests and the arithmetic mean was calculated. A matrix absolute difference algorithm is used to compare the extracted parameter array with the read data document to calculate the absolute difference of each error. A multi-dimensional floating-point matrix subtraction operation instruction is executed to remove the deviation variable and a nested absolute value calculation underlying operator is used to generate a non-negative deviation numerical matrix. A Boolean mask index threshold filtering algorithm is used to filter and extract compliant discrete sequences corresponding to the limited error range, and a preset tolerance error upper limit threshold parameter value of 0.1 is imported. Using 5 dB as the judgment threshold, the preset tolerance error upper limit threshold parameter value refers to the maximum error tolerance empirical index extracted from the previous 500 full-load operation records of the system, and a safety factor of 0.95 is assigned to the product to obtain the hard limit boundary. The absolute difference of multiple errors is compared with the judgment threshold to eliminate the physical memory index location information corresponding to the nodes that exceed the limit. The original record entries associated with the threshold detection Boolean truth judgment label are retained to extract multiple compliant discrete sequences. The horizontal one-dimensional array splicing and merging instruction is used to splice the distribution nodes of each compliant discrete sequence along the zero dimension direction. Multiple isolated data segments are connected according to the continuous memory addresses to form an integrated continuous data storage structure. The output shape of all spatial coordinate points that meet the constraints is mapped and extracted to obtain the thinned curvature array.

[0023] Generative Adversarial Network (GAN) is used to extract multivariate damping groove spatial coordinate feature fragments from the baseline parent parameters. A crossover probability threshold parameter is set, and the randomly generated values ​​are compared with the crossover probability threshold parameter. The coordinate data fragments associated with the baseline parent parameters are swapped to obtain the initial child coordinate sequence. The multidimensional spatial parameters of the initial child coordinate sequence are extracted, and a mutation probability threshold parameter is set. The randomly generated values ​​are compared with the mutation probability threshold parameter, and a perturbation numerical bias term is applied to the multidimensional spatial parameters. The mutation perturbation values ​​are superimposed to derive the child parameters.

[0024] Extract the parameters of each sequence node to the edge sharpening region space, extract the parameters of multiple thickness discrete sequence nodes, set the origin value of the physical boundary coordinates of the edge sharpening region, extract the spatial step span value, divide multiple sharpening 3D spatial meshes, extract multiple mesh position index parameters, extract multiple thickness discrete sequence associated sequence indexes, compare multiple associated sequence indexes with multiple mesh position index parameters, match the corresponding coordinate nodes of multiple spatial meshes, and project the parameters of multiple thickness discrete sequence nodes to multiple corresponding coordinate nodes.

[0025] The spatial array arrangement module includes: Parametric sorting submodule: Based on the thinned curvature array, extract the spatial coordinate data of the damping groove, calculate the coordinate distance difference parameter, summarize the difference calculation results and sort them in descending order, extract the parent features from the top sorted coordinate data, and generate the benchmark parent parameter; Perturbation-derived sub-module: Based on the baseline parent parameters, a genetic algorithm is used to extract feature fragments and perform crossover, superimpose mutation perturbation values ​​to derive sub-parameters, iteratively calculate the sub-parameters to the limit point, extract the extreme value coordinate data corresponding to the calculation limit point, and establish an extreme value coordinate set. Coordinate mapping submodule: Based on the extreme value coordinate set, it projects the node distribution values ​​to the grid system, matches the grid entity boundary constraint parameters, compares the constraint parameters to eliminate spatially conflicting node values, summarizes and retains compliant node parameters, and outputs the topology node coordinate system; The parametric sorting submodule, based on a thinned curvature array, uses the Euclidean spatial distance metric algorithm to extract the spatial coordinate data of the damping groove. It calls the underlying 3D spatial vector norm calculation package, inputting a 64-bit floating-point array matrix of multidimensional spatial coordinate points. The spatial dimension parameter value of 3 represents the physical space of the 3D Cartesian coordinate system. It iterates through the array matrix item by item to extract the 3D coordinate values ​​of individual spatial nodes, executes vector subtraction operations to obtain the spatial relative displacement vector, calculates the 2-norm value of the relative displacement vector, and outputs a multi-coordinate distance difference parameter. It then uses a 1D array stacking function to merge the multi-coordinate distance difference parameters to construct a high-dimensional distance mapping feature tensor, and uses quicksort to recursively sort the data. The underlying algorithm summarizes the difference calculation results and performs a descending sort operation. It extracts the preset pivot selection strategy parameter value 1, which refers to the system's fixed retrieval of the physical index position of the middle of the array as the comparison benchmark value. It divides the bidirectional cursor pointer and traverses the left and right sub-data sequences in a loop. It performs recursive comparison of memory values ​​and swaps physical storage address data blocks to generate a monotonically decreasing distance value sequence. It calls the high-dimensional array slice extraction built-in instructions to extract the first to 50th consecutive memory block segments after sorting, extracts the sorted coordinate data of the first column, summarizes the parent features, and assembles multiple parent features into a preset multi-dimensional tensor container structure to generate the benchmark parent parameters. The perturbation-derived submodule: Based on the baseline parent parameters, it uses a real-number encoded genetic algorithm to extract multiple feature fragments and perform crossover operations. It calls the heuristic search core computing framework to import multiple parent spatial coordinate sequence data into the initial population container, setting the crossover probability threshold parameter to 0.85 and the population size parameter to 200. It calls the underlying hardware pseudo-random number generator to generate a uniformly distributed random floating-point matrix between 0 and 1. It compares the multiple random floating-point values ​​with the crossover probability threshold parameter value. For sequences exceeding the threshold boundary, it performs multi-point crossover data bit swapping operations to generate recombinant feature sequences. It extracts a preset mutation probability parameter of 0.05, where the preset mutation probability parameter value refers to the system's initial configuration file based on the previous... The convergence rate curve of the first 1000 evolution experiments was calculated and written into the read-only entry of the system registry as a constant floating-point value. The non-uniform perturbation Gaussian mutation algorithm was used to superimpose the mutation perturbation value to derive the offspring parameters. Based on the mutation probability, a Gaussian white noise bias tensor with a mathematical expectation of 0 and a standard deviation of 0.1 was applied to the recombination feature sequence to generate multiple offspring parameters. The maximum iteration generation limit parameter was set to 500. The offspring parameters were calculated cyclically until the limit point was reached. The absolute value of the change in the floating-point value of the fitness function output was monitored to be less than 0.001 for 50 consecutive generations to confirm that the convergence limit endpoint had been reached. The extreme value coordinate data corresponding to the calculation limit point was extracted according to the memory mapping table index. Multiple two-dimensional plane and three-dimensional space extreme value coordinate data points were aggregated to form a high-dimensional floating-point tensor to establish the extreme value coordinate set. The coordinate mapping submodule, based on the extreme value coordinate set, uses a multi-dimensional spatial rasterization projection algorithm to project the node distribution values ​​to the mesh system. It extracts the three-dimensional dimensions of the physical assembly boundary of the ventilation barrier, constructs a hexahedral bounding box, sets the absolute origin of the three-dimensional global coordinate system, and calls the discretization spatial segmentation function command. It inputs a spatial step size span parameter of 5 mm to divide the system into multiple uniform orthogonal three-dimensional mesh cells. It extracts the three-dimensional floating-point sequence of extreme value coordinates, uses the round-down operator to convert the floating-point coordinates to integer mesh index numbers, and completes the projection operation. It employs a spatial boundary collision detection algorithm to match the boundary constraint parameters of the mesh entities and reads a preset interference collision threshold parameter of 2 mm. The numerical values ​​refer to the constant values ​​extracted from the underlying form fields of the equipment assembly tolerance standard database and assigned to the global variable memory space. The three-dimensional bounding box intersection test operator is used to compare the absolute difference between the coordinates corresponding to multiple mesh indices and the boundary coordinates of the entity structure to output multiple constraint parameters. Boolean mask logic is used to judge the constraint parameters and eliminate spatial conflict node values. For the tensor memory position of the parameter value where the absolute difference is less than the preset interference collision threshold, a Boolean false value label is assigned and a memory clearing instruction is triggered to eliminate the corresponding value. The Boolean true value label is extracted and associated with the original tensor memory data block to summarize and retain the compliant node parameters. Multiple compliant node parameters are merged to construct a multi-channel structured data output and build a topological node coordinate system.

[0026] The genetic algorithm extracts multiple spatial coordinate feature segments using baseline parent parameters, sets a crossover probability threshold parameter, compares the randomly generated values ​​with the crossover probability threshold parameter, extracts the corresponding position values ​​of the spatial coordinate feature segments, swaps the sequence parameters within the spatial coordinate feature segments, merges the swapped multiple sequence parameters, and obtains the initial offspring coordinate sequence.

[0027] The coupled sound field calculation module includes: Parametric overlay submodule: Based on the thinned curvature array and topological node coordinate system, extract node distribution parameters and thickness parameters, calculate the reciprocal parameter of node distance and establish cell connectivity edge weights, extract patch loss factors, merge factor parameters and edge weights to perform overlay algebra operations, and generate a set of node connectivity weights. Feature aggregation submodule: Based on the node connectivity weight set, it aggregates the spatial distribution feature parameters of nodes, performs feature-level dimensionality reduction mapping operation, extracts the corresponding measured values ​​of discrete frequency band test points and summarizes the measured values ​​of the points to form a sound pressure attenuation matrix; The parametric overlay submodule, based on a thinned curvature array and a topological node coordinate system, extracts multi-dimensional node distribution and thickness parameters using tensor slicing extraction commands. It inputs a multi-dimensional coordinate floating-point array to a spatial distance metric operator, employs a K-nearest neighbor graph construction algorithm to calculate multiple spatial node absolute displacement vectors, sets the nearest neighbor search quantity parameter to 8, extracts a set of Euclidean distance scalars, calls the underlying matrix reciprocal calculation function to compute multiple distance scalars, outputs the node distance reciprocal parameters, configures threshold filtering conditions to remove infinite singularity values, and uses a graph theory adjacency matrix construction function to assemble multiple reciprocal parameters and establish cell connections. The edge weights are calculated by calling the database read-only interface to extract the preset patch reference damping coefficient of 0.05 as the patch loss factor. The preset patch reference damping coefficient refers to the fixed value of the polyurethane damping room temperature test record automatically retrieved from the acoustic material physical property read-only form when the system starts and written into the memory constant area. The Hadamard matrix multiplication low-level operator is used to merge the multi-factor parameters and multi-edge weights to perform element-wise superposition algebraic operations. The data type is assigned as 64-bit floating point. The high-dimensional array concatenation and splicing instruction is used to merge the multi-dimensional floating point calculation result data sequence along the main diagonal to generate the node connectivity weight set. The feature aggregation submodule, based on the node connectivity weight set, uses a multidimensional tensor splicing and fusion operator to aggregate multiple node spatial distribution feature parameters. The high-dimensional connectivity weight matrix is ​​input to the core computational framework's dimensionality reduction module. Principal component analysis (PCA) is used to perform feature-level dimensionality reduction mapping operations. The parameter for retaining the principal component variance contribution rate is set to 0.95. Singular value decomposition (SVD) is used to extract the internal covariance matrix of the high-dimensional feature tensor. Multiple eigenvalues ​​and corresponding eigenvectors are solved. The top 15 largest eigenvalues ​​are extracted and associated with eigenvectors to construct a projection transformation matrix. Matrix multiplication is then performed to project the original high-dimensional feature data to a low-dimensional linear model. In the subspace, a preset frequency band cutoff limit value of 500 Hz is introduced as the extraction standard. The preset frequency band cutoff limit value refers to the low-mid frequency boundary limit value extracted by the system from the local solidified acoustic environment test standard file directory and assigned as a constant integer value in the global variable memory space. The Boolean mask frequency filtering algorithm is used to extract multiple discrete frequency band test points corresponding to the values ​​below the limit value. The tensor slicing instruction is called to read the corresponding measured values ​​of the points. The multi-channel data stacking function is used to summarize the measured values ​​of the points in row dimension priority order. The multi-dimensional floating-point data sequence is assembled into a preset two-dimensional tensor container structure to form a sound pressure attenuation matrix.

[0028] The performance degradation simulation module includes: Vector feature splicing submodule: Based on the sound pressure attenuation matrix, extract the airflow sequence and the windward temperature gradient sequence, merge the two parameters to construct an environmental feature vector, splice the vector with the time node parameters, and establish an evolution state feature set; Loss difference calculation submodule: Based on the evolution state feature set, calculates the absolute loss difference between the attenuation and calibration curve values, compares the loss difference with the system warning limit parameter and extracts the time series data corresponding to the exceedance limit to obtain the failure time calibration value. The vector feature splicing submodule, based on the sound pressure attenuation matrix, uses a multi-dimensional array physical memory slicing extraction instruction to call the read-only interface of the underlying wind speed sensor data to read the corresponding flow velocity values ​​of 1000 consecutive sampling cycles from the hardware register to extract the airflow sequence. It also calls the thermocouple array digital-to-analog converter interface to read voltage perturbation values ​​and uses linear mapping matrix operations to extract the windward temperature gradient sequence. A multi-channel sequence synchronization time axis alignment algorithm is used to match the physical timestamps of two sequences. A preset sampling frequency reference value of 100 Hz is imported. This preset sampling frequency reference value refers to a specific constant integer value read from the motherboard clock generator's underlying hardware configuration file and written to the system's global memory constant area during system startup initialization. A one-dimensional linear polynomial interpolation underlying operator is used to fill in the missing physical nodes of the discrete timestamps of the temperature gradient, and the calculation is configured. The step size parameter is set to 10 milliseconds. The output is a floating-point array structure of equal length. The high-dimensional tensor vertical stacking and merging instruction is called to merge and align the last two parameters according to the column-first physical memory arrangement order to construct a multi-channel 64-bit double-precision floating-point environmental feature vector. The system real-time clock module is called to extract the underlying high-precision timestamp data. The strong-type numerical conversion operator is used to convert the time series values ​​to double-precision floating-point time node parameters. The N-dimensional array deep splicing and fusion operator is used to splice the environmental feature vector and multiple time node parameters along the horizontal zero axis. The data storage format of the output result is assigned as a dense tensor distributed in continuous memory blocks. The shape parameter of the aggregated tensor physical dimension is configured to be a 1000x3 two-dimensional feature matrix array. The input splicing generates structured data to the system's preset dynamically expanded physical memory pool block to establish an evolution state feature set. The loss difference calculation submodule, based on the evolution state feature set, uses multidimensional tensor element-wise subtraction and absolute value nested low-level operators to read the double-precision floating-point sequence of the feature matrix memory block. It then calls the system's underlying database read-only physical storage area to retrieve the preset initial sound pressure benchmark calibration curve one-dimensional tensor. This preset initial sound pressure benchmark calibration curve one-dimensional tensor refers to the frequency response attenuation curve recorded during 24 hours of full-load operation at the factory test stage and written to the device's built-in read-only memory constant data array. Multiple evolution feature array data and benchmark calibration curve array data are input to the core central processing unit's floating-point algebra unit. The module executes matrix subtraction operations node-wise and applies absolute value mapping logic to calculate the absolute loss difference between the double-ended attenuation and the calibration curve value. A Boolean mask logic parallel comparison algorithm is used to compare and extract multiple absolute loss differences and system warning boundary parameters. A preset warning boundary parameter value of 3 dB is imported. This preset warning boundary parameter value refers to... The system backend retrieves the record set of critical points of limit performance degradation from 500 previous destructive acoustic aging fatigue physical experiments, calculates the arithmetic mean of all mathematical expectations, and then saves it to a specific floating-point constant in the operating system registry. Using a low-level tensor logic judgment operator, it iterates and compares each floating-point number within the test difference tensor with the constant limit threshold. For tensor physical memory mapping grids containing values ​​greater than or equal to 3 dB, it assigns a Boolean truth value hardware judgment identifier. It then calls a Boolean index mask physical extraction instruction to read the original tensor row index information corresponding to the Boolean truth value identifier, extracts multiple continuous time series data corresponding to the physical index associated with the out-of-bounds judgment identifier, loads the aforementioned out-of-bounds timestamps to establish a high-dimensional floating-point one-dimensional time series array, and uses a floor function to convert multiple microsecond-level timestamps to standard operating hour-cycle integer values. These values ​​are then sorted in ascending order, merged, and output to generate the failure time calibration value.

[0029] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. An array-type acoustic black hole ventilation and sound insulation barrier and its application system, characterized in that, The system includes: Boundary configuration generation module: By measuring the length and width benchmarks of the outer frame and the constant of the louver assembly spacing, the module splices the values ​​to construct a spatial position array, calculates the element product to output the volume parameter and compares the parameter with the volume upper limit, removes out-of-limit combinations and summarizes the remaining items to form the boundary dimension feature set. Contour thickness extrapolation module: Based on the boundary dimension feature set, a generative adversarial network is used to extract frequency band values ​​and perform multiply-accumulate operations in conjunction with noise to output a discrete thickness sequence. After mapping the sequence to the sharpening area to calculate the frequency response matrix, the constrained error sequence is extracted and spliced ​​to form a thinning curvature array. Spatial array arrangement module: Based on the thinned curvature array, the damping groove coordinates are extracted and the distance is calculated and arranged in descending order. The first column is taken as the parent feature. The genetic algorithm is used for crossover and mutation to derive the offspring and calculate to the limit point. The extreme value coordinates are projected to the ventilation opening and the topological node coordinate system is output. Coupled sound field calculation module: Based on the thinned curvature array and the topological node coordinate system, extract the nodes and thickness, calculate the reciprocal to construct the edge weights, extract the patch loss factor and then merge the parameters to perform superposition, aggregate the features to perform dimensionality reduction mapping and extract the test point values ​​to form a sound pressure attenuation matrix; Performance degradation prediction module: Based on the sound pressure attenuation matrix, the airflow and temperature gradient sequences are extracted and merged to construct an environmental feature vector. The vector is spliced ​​with the time state parameters, the curve loss difference is calculated and compared with the warning limit. The over-limit time series is extracted to obtain the failure time calibration value.

2. The array-type acoustic black hole ventilation and sound insulation barrier and its application system according to claim 1, characterized in that, The boundary dimension feature set includes multiple outer frame measurement length and width benchmarks, multiple louver assembly spacing constants, and multiple compliant dimension retention items. The thinning curvature array includes multiple thickness discrete sequences, multiple constrained error sequences, and multiple compliant discrete sequence distribution node combinations. The topology node coordinate system specifically includes multiple damping groove coordinates, multiple extreme value coordinates, and multiple compliant node parameters. The sound pressure attenuation matrix includes multiple node connectivity weight sets, multiple patch loss factors, and multiple discrete frequency band test point measurement values. The failure time calibration value specifically includes multiple environmental feature vectors, multiple curve loss differences, and multiple exceedance time series.

3. The array-type acoustic black hole ventilation and sound insulation barrier and its application system according to claim 1, characterized in that, The boundary configuration generation module includes: Array construction submodule: Based on the outer frame measurement length and width benchmark and the louver assembly spacing constant, extract data, align and stitch to construct coordinate nodes, accumulate span values ​​to arrange spatial array, calculate the product of array elements to output parameters, and generate spatial volume matrix; Parameter identification submodule: Based on the spatial volume matrix, read the matrix sequence to extract individual volume parameters, compare the parameter with the upper limit value of the volume, eliminate the size combination corresponding to the out-of-bounds value, reorganize the compliant size summary sequence, and generate the boundary dimension feature set.

4. The array-type acoustic black hole ventilation and sound insulation barrier and its application system according to claim 1, characterized in that, The contour thickness estimation module includes: Dimensional inference submodule: Based on the boundary dimension feature set, a generative adversarial network is used to extract multiple frequency band constraint values, and a multiply-accumulate algebraic operations are performed in conjunction with a randomly distributed noise sequence to sort out and output the thickness distribution parameters of each node section, summarize and integrate the multi-dimensional thickness distribution values, and generate a discrete thickness sequence. Response calculation submodule: Based on the thickness discrete sequence, extract and map the parameters of each sequence node to the edge sharpening region space, calculate and output the normal impedance response value of the physical surface region, integrate and reorganize the multi-band calculation parameters to construct a feature distribution combination, and establish the edge frequency response matrix; Error splicing submodule: Based on the edge frequency response matrix, extract and compare the parameter distribution of the benchmark physical sample, calculate the absolute difference of each error, screen and extract the compliant discrete sequence corresponding to the limited error range, splice the distribution node combination of each compliant discrete sequence, and obtain the thinned curvature array.

5. The array-type acoustic black hole ventilation and sound insulation barrier and its application system according to claim 1, characterized in that, The generative adversarial network extracts multiple damping groove spatial coordinate feature segments from the reference parent parameters, sets a crossover probability threshold parameter, compares the randomly generated values ​​with the crossover probability threshold parameter, swaps the coordinate data segments associated with the reference parent parameters, obtains the initial child coordinate sequence, extracts the multidimensional spatial parameters of the initial child coordinate sequence, sets a mutation probability threshold parameter, compares the randomly generated values ​​with the mutation probability threshold parameter, applies a perturbation numerical bias term to the multidimensional spatial parameters, superimposes the mutation perturbation values, and derives the child parameters.

6. The array-type acoustic black hole ventilation and sound insulation barrier and its application system according to claim 4, characterized in that, The process involves extracting and mapping the sequence node parameters to the edge sharpening region space, extracting multiple thickness discrete sequence node parameters, setting the origin value of the physical boundary coordinates of the edge sharpening region, extracting the spatial step span value, dividing multiple sharpening three-dimensional spatial meshes, extracting multiple mesh position index parameters, extracting multiple thickness discrete sequence associated sequence indices, comparing multiple associated sequence indices with multiple mesh position index parameters, matching the corresponding coordinate nodes of multiple spatial meshes, and projecting multiple thickness discrete sequence node parameters to multiple corresponding coordinate nodes.

7. The array-type acoustic black hole ventilation and sound insulation barrier and its application system according to claim 1, characterized in that, The spatial array arrangement module includes: Parametric sorting submodule: Based on the thinned curvature array, extract the spatial coordinate data of the damping groove, calculate the coordinate distance difference parameter, summarize the difference calculation results and sort them in descending order, extract the parent features from the top sorted coordinate data, and generate the benchmark parent parameter. Perturbation-derived sub-module: Based on the baseline parent parameters, a genetic algorithm is used to extract feature fragments and perform crossover, superimpose mutation perturbation values ​​to derive sub-parameters, iteratively calculate the sub-parameters to the limit point, extract the extreme value coordinate data corresponding to the calculation limit point, and establish an extreme value coordinate set. Coordinate mapping submodule: Based on the extreme value coordinate set, it projects the node distribution values ​​to the grid system, matches the grid entity boundary constraint parameters, compares the constraint parameters to eliminate spatially conflicting node values, summarizes and retains compliant node parameters, and outputs the topology node coordinate system.

8. The array-type acoustic black hole ventilation and sound insulation barrier and its application system according to claim 1, characterized in that, The genetic algorithm extracts multiple spatial coordinate feature segments using baseline parent parameters, sets a crossover probability threshold parameter, compares the randomly generated values ​​with the crossover probability threshold parameter, extracts the corresponding position values ​​of the spatial coordinate feature segments, swaps the sequence parameters within the spatial coordinate feature segments, merges the swapped multiple sequence parameters, and obtains the initial offspring coordinate sequence.

9. The array-type acoustic black hole ventilation and sound insulation barrier and its application system according to claim 1, characterized in that, The coupled acoustic field calculation module includes: Parametric overlay submodule: Based on the thinned curvature array and the topological node coordinate system, extract the node distribution parameters and thickness parameters, calculate the reciprocal parameter of node distance and establish cell connectivity edge weights, extract the patch loss factor, merge the factor parameters and edge weights to perform overlay algebra operations, and generate a set of node connectivity weights. Feature aggregation submodule: Based on the node connectivity weight set, it aggregates the spatial distribution feature parameters of the nodes, performs feature-level dimensionality reduction mapping operation, extracts the corresponding measured values ​​of discrete frequency band test points and summarizes the measured values ​​of the points to form a sound pressure attenuation matrix.

10. The array-type acoustic black hole ventilation and sound insulation barrier and its application system according to claim 1, characterized in that, The performance degradation simulation module includes: Vector feature splicing submodule: Based on the sound pressure attenuation matrix, extract the airflow sequence and the windward temperature gradient sequence, merge the two parameters to construct an environmental feature vector, splice the vector with the time node parameter, and establish an evolution state feature set; Loss difference calculation submodule: Based on the evolution state feature set, calculate the absolute loss difference between the attenuation and calibration curve values, compare the loss difference with the system warning limit parameter and extract the time series data corresponding to the exceedance limit to obtain the failure time calibration value.

Images