Artificial intelligence-based filter bag intelligent design method and system

Abstract

The application relates to the technical field of filter bag design, and discloses a filter bag intelligent design method and system based on artificial intelligence. The method comprises the following steps: obtaining a pollutant characteristic digital fingerprint library through dynamic capture of pollutant particles; obtaining a multi-dimensional material performance matrix by measuring the micro-permeability of materials; modeling fold parameters based on the matrix to obtain a self-balancing fold structure parameter group; optimizing material layer configuration based on the parameter group to form a multi-layer composite filter structure configuration table; obtaining a full-working-condition performance curve set through simulation testing; and finally accurately converting the filter bag design scheme into a producible filter bag design scheme. Based on the complex indoor air pollutant characteristics and variable environmental conditions, the application realizes accurate optimization of filter materials and structure parameters, accurately predicts the service life of the filter bag, and ensures that the design scheme can be efficiently converted into actual production, so that the filtering efficiency, service life and adaptability of the filter bag are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of filter bag design, in particular to a filter bag intelligent design method and system based on artificial intelligence. BACKGROUND

[0002] The existing filter bag design method mostly relies on traditional experience design and repeated experiments, lacking systematicness and accuracy. Design personnel usually make simple improvements or parameter adjustments based on existing products, relying on physical production and laboratory testing to verify the design effect. This method not only consumes time and effort, but also is difficult to cope with complex and variable air pollution environment. In the traditional design process, the selection of filter materials is mostly based on a single performance indicator, such as the filtration efficiency of a specific particulate matter or the initial pressure drop, and the comprehensive performance of the material under different environmental conditions is rarely considered. At the same time, the determination of the structure design of the filter bag, such as the angle, depth and density of the pleats, often relies on the personal experience of the designer, lacking a scientific optimization method. In addition, the existing technology also has great limitations in the prediction of the service life of the filter bag, and cannot accurately evaluate the influence of different environmental factors on the attenuation of filtration performance, resulting in a large difference between the actual service life and the design expectation.

[0003] However, the traditional design method has obvious shortcomings and cannot effectively cope with the current increasingly complex indoor air pollution environment. First, indoor air pollutants are diverse, including PM2.5 particulate matter, formaldehyde, VOC, etc., and their physical and chemical characteristics and distribution rules are complex and variable, making it difficult for traditional experience design to accurately optimize for such complex pollution environment. Secondly, there is a complex interaction between filter materials and structural parameters, and traditional methods cannot fully consider the coupling effect of these factors, resulting in unstable or low-efficiency performance of the designed filter bag. Thirdly, environmental factors such as temperature and humidity have a significant impact on the performance of filter materials, and the existing technology lacks systematic analysis and prediction ability of these influences, making it difficult to guarantee the performance of the filter bag in the actual use environment. Finally, the traditional design-manufacturing process lacks effective data feedback and optimization mechanism, and cannot accurately adjust the design according to the actual production situation, resulting in low conversion efficiency from design to production. SUMMARY

[0004] The present application provides a filter bag intelligent design method and system based on artificial intelligence, which is used to accurately optimize filter materials and structural parameters based on the characteristics of complex indoor air pollutants and variable environmental conditions, accurately predict the service life of the filter bag, and ensure that the design scheme can be efficiently converted into actual production, thereby improving the filtration efficiency, service life and adaptability of the filter bag.

[0005] In a first aspect, the application provides an intelligent design method for a filter bag based on artificial intelligence, which comprises: collecting multi-dimensional data of particle size distribution and adsorption characteristics of indoor air pollutants by a pollutant particle dynamic capture device to obtain a pollutant characteristic digital fingerprint library; accurately quantifying porosity, active site distribution and surface charge of a filter material by a material microscopic permeability measuring instrument according to the pollutant characteristic digital fingerprint library to obtain a multi-dimensional material performance matrix; parameterizing modeling and airflow distribution calculation of a filter bag wrinkle angle, depth and density based on the multi-dimensional material performance matrix to obtain a self-balancing wrinkle structure parameter set; optimizing and adjusting the combination order, thickness ratio and contact interface of different material layers of the filter bag according to the self-balancing wrinkle structure parameter set to obtain a multi-layer composite filter structure configuration table; simulating and testing the filtration efficiency and service life of the filter bag in a high-humidity and high-concentration pollutant environment by using the multi-layer composite filter structure configuration table to obtain a full-working-condition performance curve set; and accurately converting and checking a filter bag material formula, structure size and production process parameters according to the full-working-condition performance curve set to obtain a filter bag design scheme that can be directly used for production.

[0006] In a second aspect, the application provides an intelligent design system for a filter bag based on artificial intelligence, which comprises:

[0007] A collecting module for collecting multi-dimensional data of particle size distribution and adsorption characteristics of indoor air pollutants by a pollutant particle dynamic capture device to obtain a pollutant characteristic digital fingerprint library;

[0008] A quantifying module for accurately quantifying porosity, active site distribution and surface charge of a filter material by a material microscopic permeability measuring instrument according to the pollutant characteristic digital fingerprint library to obtain a multi-dimensional material performance matrix;

[0009] A calculating module for parameterizing modeling and airflow distribution calculation of a filter bag wrinkle angle, depth and density based on the multi-dimensional material performance matrix to obtain a self-balancing wrinkle structure parameter set;

[0010] An adjusting module for optimizing and adjusting the combination order, thickness ratio and contact interface of different material layers of the filter bag according to the self-balancing wrinkle structure parameter set to obtain a multi-layer composite filter structure configuration table;

[0011] A testing module for simulating and testing the filtration efficiency and service life of the filter bag in a high-humidity and high-concentration pollutant environment by using the multi-layer composite filter structure configuration table to obtain a full-working-condition performance curve set;

[0012] The check module is used for accurately converting and checking the filter bag material formula, structure size and production process parameters according to the full working condition performance curve set, so as to obtain a filter bag design scheme which can be directly used for production.

[0013] The third aspect of the present application provides a computer device, comprising: a memory and at least one processor, the memory has instructions stored therein; the at least one processor invokes the instructions in the memory to enable the computer device to execute the above-mentioned artificial intelligence-based filter bag intelligent design method.

[0014] The fourth aspect of the present application provides a computer readable storage medium, the computer readable storage medium has instructions stored therein, when the instructions are run on a computer, the computer is enabled to execute the above-mentioned artificial intelligence-based filter bag intelligent design method.

[0015] The technical scheme provided in the application, through the multi-dimensional data acquisition of indoor air pollutants by the pollutant particle dynamic capture device, an accurate pollutant characteristic digital fingerprint library is established, the fingerprint library contains key information such as the particle size distribution and adsorption characteristics of the pollutants, and provides accurate data basis for subsequent design, solving the problem of insufficient pollutant characteristic cognition in traditional design. Secondly, the filtration material is accurately quantified by using the material micro-penetration measuring instrument, and a multi-dimensional material performance matrix is generated, which comprehensively represents the characteristics of the material such as porosity, active site distribution and surface charge, so that the designer can more accurately select the filtration material suitable for a specific pollutant. Thirdly, based on the parameterized modeling and airflow distribution calculation of the multi-dimensional material performance matrix, a self-balancing pleat structure parameter group is generated, and the parameters such as the pleat angle, depth and density are optimized by the artificial intelligence algorithm, so as to realize the optimal design of the filter bag structure, and significantly improve the filtration efficiency and service life. In addition, the combination order, thickness ratio and contact interface of different material layers of the filter bag are optimized and adjusted to form a multi-layer composite filter structure configuration table, which fully plays the synergistic effect of each material layer and enhances the overall performance of the filtration system. Especially worth mentioning is that the virtual test and life prediction are carried out by using the artificial intelligence algorithm, the performance curve set under all working conditions is generated by simulating the performance of the filter bag in the high-humidity and high-concentration pollutant environment, the performance and service life of the product under various environmental conditions are accurately predicted, and the limitation of relying on physical test in the traditional method is avoided. Finally, the accurate conversion and verification are carried out according to the performance curve set under all working conditions, and the filter bag design scheme directly used for production is generated, realizing the seamless connection from design to production, improving the design conversion efficiency and product consistency. The application of artificial intelligence algorithm in the application, especially the deep learning and optimization algorithm in the aspects of pollutant data fusion processing, material performance prediction, structure parameter optimization and life prediction, greatly improves the intelligent level and accuracy of the design process, so that the filter bag design is no longer dependent on experience and trial and error, but based on accurate data and scientific calculation, realizing the personalized and accurate design for complex indoor air pollution environment, thereby significantly improving the filtration efficiency, service life and environmental adaptability of the filter bag. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are some embodiments of the application, and other drawings can be obtained by those skilled in the art without creating laborious work on the basis of these drawings.

[0017] Figure 1 An embodiment schematic diagram of the intelligent filter bag design method based on artificial intelligence in the embodiments of the application;

[0018] Figure 2 An embodiment of the filter bag intelligent design system based on artificial intelligence in the embodiments of the present application is shown in the figure;

[0019] Figure 3 An embodiment of the computer device in the embodiments of the present application is shown in the figure. DETAILED DESCRIPTION

[0020] The embodiments of the present application provide a filter bag intelligent design method and system based on artificial intelligence. The terms "first", "second", "third", "fourth" and the like (if any) in the specification and claims of the present application and the above-mentioned figures are used to distinguish similar objects, and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" or "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not necessarily limit to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0021] For the sake of understanding, the specific process of the embodiments of the present application is described below. Please refer to Figure 1 An embodiment of the filter bag intelligent design method based on artificial intelligence in the embodiments of the present application includes:

[0022] Step S101, multi-dimensional data acquisition of particle size distribution and adsorption characteristics of indoor air pollutants is performed by a pollutant particle dynamic capture device, and a pollutant characteristic digital fingerprint library is obtained;

[0023] Step S102, according to the pollutant characteristic digital fingerprint library, the porosity, active site distribution and surface charge of the filter material are accurately quantified by using a material microscopic permeability measuring instrument, and a multi-dimensional material performance matrix is obtained;

[0024] Step S103, based on the multi-dimensional material performance matrix, the filter bag pleat angle, depth and density are parameterized modeling and airflow distribution calculation, and a self-balancing pleat structure parameter group is obtained;

[0025] Step S104, according to the self-balancing pleat structure parameter group, the combination order, thickness ratio and contact interface of different material layers of the filter bag are optimized and adjusted, and a multi-layer composite filter structure configuration table is obtained;

[0026] Step S105, using the multi-layer composite filter structure configuration table, the filter efficiency and service life of the filter bag in a high humidity and high concentration pollutant environment are simulated and tested, and a full working condition performance curve set is obtained;

[0027] Step S106, according to the full working condition performance curve set, the filter bag material formula, structure size, production process parameters are accurately converted and checked, and the filter bag design scheme directly used for production is obtained.

[0028] It can be understood that the execution subject of the present application can be an intelligent filter bag design system based on artificial intelligence, and can also be a terminal or a server, which is not limited here. The server is taken as an example for illustration.

[0029] Specifically, the PM2.5 particle concentration is monitored in real time using a high-precision light scattering sensor. The sensor is based on Mie scattering principle. When the laser beam irradiates the particulate matter in the air, it will produce scattered light. The sensor detects the scattered light intensity and converts it into an electrical signal. The particulate matter concentration is calculated by an algorithm. At the same time, the gas chromatograph analyzer measures the concentration gradient of gaseous pollutants such as formaldehyde and VOC. The gas chromatograph analyzer separates the mixed gas through a chromatographic column, and then identifies and quantifies the concentration of different pollutants by a detector. The electrostatic adsorption test board is used to quantify and record the surface adhesion of pollutants under different temperature and humidity conditions. A specific voltage is applied to the surface of the test board, and the adsorption rate and strength of the pollutants are recorded. The data fusion processor correlates and analyzes these three kinds of data, generates a time-space distribution characteristic map of pollutants through time stamp alignment and spatial interpolation algorithm, and extracts key pollutant characteristic parameters by principal component analysis algorithm to establish a pollutant behavior mode library. Finally, through cross-validation and normalization processing, a pollutant characteristic digital fingerprint library is formed, which contains multi-dimensional information such as pollutant particle size, concentration and adsorption characteristics.

[0030] According to the pollutant characteristic digital fingerprint library, the specific quantification is carried out by using the material micro-penetration measuring instrument. The specific surface area of the filter material is measured by nitrogen adsorption method. The nitrogen adsorption method is based on BET theory. The amount of nitrogen adsorbed on the surface of the material is measured at low temperature, so as to calculate the specific surface area and pore size distribution of the material. The electron microscopic scanning technology is used to image the surface morphology of the filter material at high magnification, so as to obtain the material surface microstructure graph with nanometer resolution, and display the fiber arrangement, pore distribution and surface roughness of the material surface. The adsorption capacity of the filter material is tested by the thermogravimetric analyzer under different temperature conditions. The mass change of the material during temperature change is recorded, and the adsorption thermodynamic characteristics of the material are calculated. The material pore size distribution curve and surface microstructure graph are fused by image recognition algorithm and three-dimensional reconstruction technology to establish the three-dimensional pore network topology structure of the material. The fluorescence labeling method is used to combine specific fluorescent dyes with active sites of the material, and the spatial distribution of the active sites is observed and quantified under the fluorescence microscope. The correlation analysis algorithm combines the active site spatial distribution graph with the adsorption thermodynamic characteristic graph to construct a capture efficiency prediction model, and finally generates a multi-dimensional material performance matrix.

[0031] When parameterizing the pleat of the filter bag based on the multi-dimensional material performance matrix, the geometric parameter generator first discretizes the pleat structure, decomposes the continuous pleat form into a finite number of grid points, and forms a grid point set of the pleat unit. The physical parameters in the multi-dimensional material performance matrix, such as elastic modulus, porosity, and surface energy, are mapped to the grid points to construct a digital pleat material model. The flow field boundary conditions include inlet flow rate, outlet pressure, and side wall no-slip condition. The pressure drop and airflow flux distribution of each pleat unit are calculated by finite element analysis. The pressure imbalance area is adjusted by the gradient descent algorithm to generate an angle optimization data table. The response surface analysis method is combined with the pleat depth variable to establish a mathematical relationship between the depth and the density, forming a pleat geometric coupling relationship diagram. Finally, the parameters are continuously adjusted by the iterative optimization algorithm until the airflow distribution uniformity reaches the preset threshold, and the self-balancing pleat structure parameter set is generated.

[0032] When optimizing and adjusting different material layers of the filter bag according to the self-balancing pleat structure parameter set, the material compatibility analyzer first measures the interfacial force between candidate filter materials, and obtains the material interface bonding strength data through tensile testing and shear testing. The capture capacity complementarity evaluation is based on the data in the pollutant characteristic digital fingerprint library, calculates the capture efficiency of different materials for various pollutants, and forms a material complementarity score table. The exhaustive analysis algorithm systematically evaluates the arrangement and combination schemes of different material layers, and generates a material ordering scheme library through directional filtration coefficient calculation. The interlayer pressure drop coordination calculation determines the optimal thickness ratio of each layer of material using a fluid mechanics model, and constructs a thickness ratio relationship matrix. The interface stress analysis identifies stress concentration problems at the contact points between material layers, and generates an interface optimization strategy table by adjusting the contact method. All these analysis results are finally integrated to form a multi-layer composite filter structure configuration table.

[0033] When simulating the test using the multi-layer composite filter structure configuration table, the environmental parameter control box sets the temperature (15-40℃), humidity (30-90%) and pollutant concentration (10-1000μg / m 3A multi-dimensional test condition matrix is constructed by different combinations of the above-mentioned factors. The digital twin prototype constructs a virtual model based on the parameters in the multi-layer composite filter structure configuration table, loads the environmental conditions in the computational fluid dynamics software, performs virtual wind tunnel testing, and obtains initial filtration resistance data. The accelerated aging process simulates the pollutant accumulation process during long-term use through a mathematical model, calculates the capture saturation curve, and draws the decay spectrum of filtration efficiency over time. The humidity response strain analysis studies the moisture absorption deformation characteristics of the material under different humidity conditions, and constructs a humidity response strain map. The life prediction analysis combines the initial filtration resistance and efficiency decay data to evaluate the impact of high-concentration pollutants on the saturation time of the material, and generates a life prediction data set. The boundary condition screening and performance interpolation calculation processes the prediction data through statistical methods to form a full-condition performance curve set covering various working conditions. According to the full-condition performance curve set, the material formula analyzer performs reverse analysis on the indicators in the performance curve to determine the material composition ratio relationship and generate a material formula list. The availability assessment checks the availability of the formula items in the existing supply chain, matches the supplier resource library, and forms a material procurement parameter table. The key structure size data is extracted from the multi-layer composite filter structure configuration table and the full-condition performance curve set to construct a manufacturing size specification card. The process adaptability analysis converts the design parameters into specific production equipment operation parameters to form a process flow instruction set. The small batch trial collects deviation data in actual production, establishes a process correction factor table, and is used to fine-tune the material formula and manufacturing size, finally integrating to form a filter bag design scheme that can be directly used for production.

[0034] For example, when the capture device detects that the average diameter of PM2.5 particles in a certain indoor environment is 0.8 microns, the concentration is 75 μg / m 3 , the formaldehyde concentration is 0.12 mg / m 3 , the VOC concentration is 1.5 mg / m 3 , and the electrostatic adsorption test records that the pollutant adhesion force is 2.3 mN / cm 2 at 45% humidity, these data are fused to form a pollutant spatiotemporal distribution feature map. Material micro-penetration measurement shows that a certain filter material has a porosity of 78%, a specific surface area of 850 m 2 / g, and an active site density of 3.2 x 1014 / cm 2 . Parameterized modeling calculates that the optimal pleat angle is 22°, the depth is 15 mm, and the density is 4.2 pleats / cm for the most uniform airflow distribution. The multi-layer composite filter structure adopts a three-layer design of electrostatic fibers, activated carbon fibers, and HEPA fibers, with a thickness ratio of 2:3:5, and the interface uses a hot pressing combined process. Simulation tests show that at 80% humidity and 250 μg / m 3Under the pollution concentration, the filtration efficiency decreases from the initial 99.5% to 90% in 1200 hours, and these data are converted into specific material formula and production process parameters through reverse analysis, and finally form a practical filter bag design scheme.

[0035] In the embodiments of the present application, the indoor air pollutants are collected by the pollutant particle dynamic capture device, and an accurate pollutant characteristic digital fingerprint library is established, which contains the particle size distribution, adsorption characteristics and other key information of the pollutants, providing accurate data basis for subsequent design and solving the problem of insufficient understanding of pollutant characteristics in traditional design. Secondly, the filtration material is accurately quantified by using the material micro-penetration measuring instrument, and a multi-dimensional material performance matrix is generated, which comprehensively represents the characteristics of the material such as porosity, active site distribution and surface charge, so that the designer can more accurately select the filtration material suitable for specific pollutants. Thirdly, based on the multi-dimensional material performance matrix, parameterized modeling and airflow distribution calculation are carried out to generate a self-balancing pleat structure parameter group, and through the optimization of parameters such as pleat angle, depth and density by artificial intelligence algorithm, the optimal design of the filter bag structure is realized, which significantly improves the filtration efficiency and service life. In addition, the combination order, thickness ratio and contact interface of different material layers of the filter bag are optimized and adjusted to form a multi-layer composite filter structure configuration table, which fully plays the synergistic effect of each material layer and enhances the overall performance of the filtration system. It is particularly worth mentioning that the virtual test and life prediction are carried out by using the artificial intelligence algorithm, the performance curve set under all working conditions is generated by simulating the performance of the filter bag in high humidity and high concentration pollutant environment, and the performance and service life of the product under various environmental conditions are accurately predicted, avoiding the limitations of relying on physical tests in traditional methods. Finally, the precise conversion and verification are carried out according to the performance curve set under all working conditions to generate a filter bag design scheme that can be directly used for production, realizing seamless connection from design to production and improving design conversion efficiency and product consistency. The application of artificial intelligence algorithm in the present application, especially the deep learning and optimization algorithm in the aspects of pollutant data fusion processing, material performance prediction, structure parameter optimization and life prediction, greatly improves the intelligent level and accuracy of the design process, so that the filter bag design is no longer dependent on experience and trial and error, but based on accurate data and scientific calculation, realizing personalized and accurate design for complex indoor air pollution environment, thereby significantly improving the filtration efficiency, service life and environmental adaptability of the filter bag.

[0036] In a specific embodiment, the process of step S101 can specifically include the following steps:

[0037] (1) Real-time concentration monitoring of PM2.5 particles in indoor air by high-precision light scattering sensor to obtain particle size distribution histogram;

[0038] (2) The concentration gradient of indoor formaldehyde and VOC gaseous pollutants is measured by a gas chromatograph analyzer to form a pollutant concentration change curve;

[0039] (3) The surface adhesion of different pollutants under various temperature and humidity conditions is quantitatively recorded by using an electrostatic adsorption test board to construct a pollutant adhesion characteristic data set;

[0040] (4) According to the obtained particle size distribution histogram, pollutant concentration change curve and pollutant adhesion characteristic data set, multi-source data correlation analysis is performed by a data fusion processor to generate a pollutant spatiotemporal distribution characteristic map;

[0041] (5) The generated pollutant spatiotemporal distribution characteristic map is imported into a feature extraction unit to extract key pollutant characteristic parameters and establish a pollutant behavior mode library;

[0042] (6) According to the established pollutant behavior mode library and historical environmental condition data, cross-validation and normalization processing are performed to form a pollutant characteristic digital fingerprint library.

[0043] Specifically, the PM2.5 particles in the indoor air are monitored in real time by a high-precision light scattering sensor to obtain a particle size distribution histogram. The high-precision light scattering sensor is a detection device based on the principle of light scattering. When a laser beam shines on the suspended particulate matter in the air, the particulate matter will scatter light, and the sensor captures these scattered light and converts the light signal into an electrical signal through a photoelectric conversion element. The signal processing circuit amplifies and filters the electrical signal, and then calculates the size and concentration of the particulate matter through the Mie scattering theory. The sensor collects data in real time, sampling 60 times per second, and after continuous monitoring for 30 minutes, 180,000 data points are collected and grouped according to the particle diameter size to form a particle size distribution histogram. The histogram has a horizontal axis representing the particle diameter range (usually divided into 50 intervals from 0.1 pm to 10 pm) and a vertical axis representing the number or concentration of particles in each interval. The concentration gradient of indoor formaldehyde and VOC gaseous pollutants is measured using a gas chromatograph analyzer to form a pollutant concentration change curve. The gas chromatograph analyzer is an instrument for separating and detecting the components of complex mixed gases. Its working principle is to use the difference in the distribution coefficient between the stationary phase and the mobile phase of different compounds to separate the components in the mixture in the chromatographic column. The analyzer is equipped with a special chromatographic column and a hydrogen flame ionization detector, which can detect ppb (parts per billion) level of formaldehyde and VOC concentration. When performing concentration gradient measurement, the analyzer collects samples at different locations in the room (usually 9 measurement points, forming a three-dimensional grid) at a preset time interval (every 10 minutes) for 3 hours, obtaining 162 concentration data points. These data are arranged in chronological order, and a continuous pollutant concentration change curve is generated by interpolation algorithm, with time on the horizontal axis and pollutant concentration on the vertical axis.

[0044] At the same time, the surface adhesion of different pollutants under various temperature and humidity conditions is quantitatively recorded using an electrostatic adsorption test board to construct a pollutant adhesion characteristic dataset. The electrostatic adsorption test board is a device specifically designed to measure the adsorption characteristics of micro-particles, composed of a metal plate with controllable electric charge on the surface, a micro-electronic force sensor, and an environmental parameter control system. During testing, the control system sets specific temperature (15℃ to 40℃, interval 5℃) and humidity (30% to 90%, interval 10%) combinations, forming 42 different environmental conditions. Under each environmental condition, the test board surface is applied with different voltages (0V to 5000V, interval 500V) to measure the adhesion of different types of pollutants (PM2.5, formaldehyde condensate, VOC condensate). The micro-electronic force sensor records the force changes during the adsorption and desorption of pollutants, calculates the surface adhesion value, and finally obtains a pollutant adhesion characteristic dataset containing 42x11x3 = 1386 data points.

[0045] According to the obtained particle size distribution histogram, pollutant concentration variation curve and pollutant adhesion characteristic data set, multi-source data correlation analysis is carried out through a data fusion processor to generate a pollutant spatiotemporal distribution characteristic map. The data fusion processor is a computing device specially used for processing multi-source heterogeneous data, and uses Kalman filtering algorithm to synchronize and align time series data. First, time stamp alignment is performed to synchronize the three kinds of data according to the collection time; then, spatial registration is performed to map the data collected at different positions to a unified three-dimensional spatial coordinate system; then, data standardization is performed to eliminate the differences caused by different dimensions; finally, the correlation between the data is extracted through a tensor decomposition algorithm, and a continuous pollutant spatiotemporal distribution characteristic map is generated using a spatial interpolation method. The characteristic map is a four-dimensional data body, three dimensions represent spatial coordinates, and one dimension represents time, and each spatiotemporal point contains multi-dimensional information such as pollutant concentration, particle size distribution and adhesion characteristic.

[0046] The generated pollutant spatiotemporal distribution characteristic map is imported into a feature extraction unit to extract key pollutant feature parameters and establish a pollutant behavior mode library. The feature extraction unit is a data processing module designed based on deep learning algorithm, which uses convolutional neural network to analyze the spatiotemporal distribution characteristic map. The unit first performs dimension reduction processing on the four-dimensional data body, extracts the main variation characteristics through principal component analysis method; then applies non-negative matrix factorization algorithm to decompose the complex pollutant distribution mode into linear combination of several basic modes; then uses clustering algorithm to classify similar pollutant behavior modes to form typical pollutant behavior categories; finally, the association rule mining algorithm is used to find the association rules between environmental conditions and pollutant behavior. After this series of processing, key feature parameters including diffusion rate, settling coefficient and adsorption saturation are extracted from massive data, and a pollutant behavior mode library containing typical pollutant behavior modes is established.

[0047] According to the established pollutant behavior mode library and historical environmental condition data, cross-validation and normalization processing are performed to form a pollutant characteristic digital fingerprint library. Cross-validation is a statistical method for evaluating the generalization ability of a model, which repeatedly trains and validates the model by dividing the data set into training set and validation set. In this step, the K-fold cross-validation method is used, the pollutant behavior mode library data is randomly divided into 10 parts, each time 9 parts are used to train the model, and the remaining 1 part is used to validate the model prediction accuracy, and the average result is taken after 10 times. Normalization processing is the process of converting data of different dimensions and different distributions to a unified scale, and Z-score standardization method is used to process each feature parameter so that the mean is 0 and the standard deviation is 1. After cross-validation, stable and reliable pollutant behavior modes are selected, and through normalization processing, the data format and scale are unified, and finally a pollutant characteristic digital fingerprint library containing multi-dimensional information such as pollutant particle size characteristics, concentration distribution characteristics, adsorption characteristics and diffusion characteristics is formed.

[0048] For example, in an office environment, a high-precision light scattering sensor continuously monitored PM2.5 particles for 30 minutes. The collected data showed that most particles were concentrated in the 0.5μm to 2.5μm range, forming a right-skewed single-peak histogram. A gas chromatograph performed gradient measurements at nine measurement points over three hours, finding a formaldehyde concentration of 0.08 mg / m³. 3 Up to 0.15 mg / m 3 The data fluctuated between different temperature and humidity levels, exhibiting a clear time-periodic variation and highly correlated with the indoor air conditioning operating cycle. Electrostatic adsorption tests under different temperature and humidity combinations revealed that when humidity increased from 30% to 70%, the surface adhesion of PM2.5 particles increased by 2.8 times, while the effect of temperature was relatively small. The data fusion processor performed correlation analysis on these three sets of data, finding that the southwest corner of the office had the highest pollutant concentration, peaking 15 minutes after the air conditioning was turned on. The feature extraction unit extracted key parameters from the spatiotemporal distribution feature map, including a pollutant diffusion rate of 3.2 cm / min and a sedimentation coefficient of 0.018 min⁻¹. Through 10-fold cross-validation and Z-score normalization, these feature parameters were compared with historical data, confirming that the pollutants in this environment mainly originated from office equipment release, forming a digital fingerprint of pollutant characteristics. This digital fingerprint data directly guides the selection of subsequent filter materials and the design of filter bag structures, enabling the final filter bag to specifically filter the main pollutants in the environment.

[0049] In one specific embodiment, the process of performing step S102 may specifically include the following steps:

[0050] (1) The specific surface area of ​​the filter material was measured by nitrogen adsorption method, and the pore size distribution curve of the material was recorded.

[0051] (2) Electron microscopy scanning technology was used to perform high-magnification imaging of the surface morphology of the filter material to obtain a microstructure map of the material surface.

[0052] (3) The adsorption capacity of the filter material under different temperature conditions was quantitatively tested using a thermogravimetric analyzer to form an adsorption thermodynamic characteristic diagram.

[0053] (4) The material pore size distribution curve and the material surface microstructure map are fused together to establish the three-dimensional pore network topology of the material.

[0054] (5) Active sites in the three-dimensional pore network topology of the material are located and their density is measured by fluorescent labeling method, and a spatial distribution map of active sites is constructed.

[0055] (6) Combine the active site spatial distribution map with the adsorption thermodynamic characteristic map, construct a material capture efficiency prediction model through correlation analysis, and generate a multi-dimensional material performance matrix.

[0056] Specifically, the specific surface area of the filter material is determined by nitrogen adsorption method, and the pore size distribution curve of the material is recorded. Nitrogen adsorption method is a classical method for determining the specific surface area and pore structure of porous materials, which is based on BET theory (Brunauer-Emmett-Teller theory). The theory describes the process of multilayer adsorption of gas molecules on the surface of a solid. During the determination, the filter material sample is placed in a sample tube, and after degassing treatment to remove surface impurities and moisture, nitrogen adsorption-desorption experiment is carried out at liquid nitrogen temperature (-196℃). The test instrument automatically controls the change of nitrogen pressure, records the amount of nitrogen adsorbed by the material at different relative pressures, and draws the adsorption-desorption isotherm. The pore size distribution curve of the material is calculated by BJH method (Barrett-Joyner-Halenda method), which shows the pore size (usually a number of discrete points in the range of 1-100 nanometers) on the horizontal axis and the pore volume or pore area distribution corresponding to the pore size on the vertical axis. At the same time, the specific surface area of the material is calculated, which is the total surface area per unit mass of the material, with the unit of square meter per gram (m 2 / g).

[0057] The surface morphology of the filter material is imaged at high magnification using electron microscopic scanning technology to obtain a microstructure image of the material surface. Electron microscopic scanning technology uses secondary electrons, backscattered electrons, and other signals generated by the interaction of an electron beam with the sample surface to collect and convert image signals through a detector, thereby enabling observation of the sample surface morphology. In filter material analysis, the sample is first cut to an appropriate size and fixed on a sample stage with conductive tape, and then treated with gold spraying or carbon spraying to enhance conductivity. The sample is then placed in the vacuum chamber of a scanning electron microscope, and the electron beam is scanned on the sample surface by controlling the electron beam acceleration voltage (usually 5-20 kV) and focusing system. The secondary electron detector collects signals, and the image processing system forms a microstructure image of the material surface. Usually, multiple images are collected at different magnifications (from 100x to 100,000x) for each sample to fully understand the material structure from macro to micro, including fiber distribution, surface roughness, pore morphology, and other information. Subsequently, the adsorption capacity of the filter material under different temperature conditions is quantitatively tested using a thermogravimetric analyzer to form an adsorption thermodynamic property diagram. A thermogravimetric analyzer is an instrument that precisely measures the mass change of a sample during temperature changes, used to study the thermal stability, adsorption-desorption behavior, and reaction kinetics of materials. During testing, the filter material sample is placed on a precision balance in a specific atmosphere (usually nitrogen or air) and heated according to a pre-set program (from room temperature to 500°C at a rate of 10°C / min). At the same time, the relationship between sample mass change and temperature is recorded, i.e., the thermogravimetric curve (TG curve). For adsorption property research of filter materials, the sample is first treated at high temperature to remove moisture and volatile substances, then cooled to room temperature, and then introduced into a gas containing the pollutants to be tested (such as formaldehyde or VOCs), and the adsorption mass increase of the material at different temperatures is recorded. Through data processing, the adsorption capacity at different temperatures is calculated, and the adsorption thermodynamic property diagram is drawn, which reflects the dependence of the material's adsorption capacity on temperature for specific pollutants.

[0058] The material pore size distribution curve and the material surface microstructure image are data fused to establish the material three-dimensional pore network topology. Data fusion is a process of integrating data of different sources and different properties to form more complete information. In this method, first, the electron microscope image is digitally processed, including image enhancement, noise removal and edge detection, to extract features such as fiber direction and pore distribution on the material surface. Then, different structural units such as fibers, cross points and pores are identified and labeled using image segmentation algorithms. At the same time, the statistical data in the pore size distribution curve are spatially mapped with the image information, and the two-dimensional surface information is expanded to a three-dimensional network structure through a structure reconstruction algorithm. In specific implementation, a Markov random field model is used to describe the spatial correlation of pores inside the material, and a Monte Carlo simulation method is used to reconstruct the three-dimensional structure based on two-dimensional slice information. The final three-dimensional pore network topology of the material is a digital model containing pore position, size, connectivity and spatial distribution, which can intuitively display the internal structure characteristics of the material.

[0059] The active sites in the three-dimensional pore network topology of the material are located and the density is measured by fluorescence labeling method to construct the active site spatial distribution map. Fluorescence labeling method is a technology that uses specific fluorescent dyes or fluorescent markers to specifically bind to target substances, and observes and analyzes the distribution of target substances under a fluorescence microscope. In the analysis of filter materials, first, a fluorescent dye that can specifically bind to the active sites of the material (such as hydroxyl, carboxyl, amino and other functional groups) is selected, and the material sample is immersed in the dye solution to allow the fluorescent molecules to bind to the active sites. After washing to remove unbound dye, three-dimensional scanning imaging is performed under a confocal fluorescence microscope. The instrument records the fluorescence signal intensity and distribution of each layer by scanning the sample layer by layer to form a three-dimensional fluorescence intensity data set. Through quantitative analysis of fluorescence intensity by image processing software, the density (number of active sites per unit volume or unit mass of material) and distribution uniformity of active sites are calculated. Combined with the three-dimensional pore network topology obtained before, the active site information is mapped to the three-dimensional structure to construct the active site spatial distribution map, accurately characterizing the position and concentration gradient of active sites in the material space.

[0060] The active site spatial distribution map is combined with the adsorption thermodynamic characteristic map, a material capture efficiency prediction model is constructed through correlation analysis, and a multi-dimensional material performance matrix is generated. Correlation analysis is a statistical method for studying the relationship between variables. In this method, the correlation coefficients between active site density, distribution characteristics and adsorption capacity of the material under different conditions are first calculated to identify the parameter pairs with significant correlation. Then, a multivariate regression analysis method is used to establish the mathematical relationship between the active site characteristic parameters and the adsorption performance, and a prediction model is constructed. The model can predict the capture efficiency of the material for different pollutants according to the structural characteristics and active site distribution of the material. Based on this model, the performance of the material under different working conditions (temperature, humidity, pollutant concentration, etc.) is systematically evaluated, and a multi-dimensional material performance matrix is generated. This matrix is a high-dimensional data structure, each dimension corresponds to a performance parameter or working condition variable, and the matrix element value represents the performance index of the material under specific conditions, such as capture efficiency, adsorption capacity, pressure drop, etc.

[0061] In a specific embodiment, the process of performing step S103 can specifically include the following steps:

[0062] (1) Discretize the initial geometric morphology of the pleat structure through the geometric parameter generator to form a pleat unit grid point set;

[0063] (2) Map the material physical parameters in the multi-dimensional material performance matrix to the pleat unit grid point set to construct a digital pleat material model;

[0064] (3) Apply flow field boundary conditions to the digital pleat material model to determine the airflow flux distribution of each pleat unit through segmented pressure drop calculation;

[0065] (4) Identify pressure imbalance areas based on the airflow flux distribution, adjust the pleat angle parameters by gradient, and generate an angle optimization data table;

[0066] (5) Combine the angle optimization data table with the pleat depth variable to determine the depth-density relationship curve through response surface analysis, and form a pleat geometric coupling relationship diagram;

[0067] (6) Perform airflow uniformity evaluation and parameter iterative optimization according to the pleat geometric coupling relationship diagram to generate a self-balancing pleat structure parameter group.

[0068] Specifically, the initial geometric morphology of the pleat structure is discretized by a geometric parameter generator to form a pleat cell grid point set. The geometric parameter generator is a computational tool specifically designed to generate digital models of complex geometric structures. For the filter bag pleat structure, parameterized modeling techniques are used to convert continuous pleat morphology into discrete mathematical representations. Discretization is the process of dividing a continuum into a finite number of discrete elements, facilitating subsequent numerical calculations and analysis. In specific implementation, first, the geometric description equation of the pleat structure is established, and the basic shape characteristics of the pleat are defined, including pleat angle, pleat depth, pleat spacing, etc. Then, finite element meshing technology is used to mesh the pleat structure, dividing the continuous pleat into thousands of small cells to form a pleat cell grid point set. The density of the grid points is dynamically adjusted according to the calculation accuracy requirements, and the grid density is usually higher at the corners of the pleat where the geometry changes dramatically. Each grid point records the three-dimensional spatial coordinates and local geometric feature parameters, such as curvature, normal vector, etc.

[0069] The material physical parameters in the multi-dimensional material performance matrix are mapped to the pleat cell grid point set to construct a digital pleat material model. The multi-dimensional material performance matrix contains various physical performance parameters of the filter material, such as porosity, permeability, elastic modulus, surface charge density, etc. Parameter mapping is a process of associating these material properties with spatial positions, assigning appropriate material characteristic values to each grid point. The mapping process uses a spatial interpolation algorithm to calculate the accurate parameter values at the micro grid points based on the distribution characteristics of the material at the macro scale. For heterogeneous materials, use layered mapping technology to assign the characteristics of different material layers to the grid points in the corresponding regions. The digital pleat material model is a computational model that integrates geometric structure and material physical properties, each grid point contains both position information and a complete set of material performance parameters.

[0070] After the digital pleat material model is established, it needs to be subjected to flow field boundary conditions to determine the gas flow flux distribution of each pleat cell through segmented pressure drop calculation. The flow field boundary conditions include inlet flow rate, outlet pressure, and side wall no-slip condition, etc. The segmented pressure drop calculation is based on the extended form of Darcy's formula to calculate the pressure gradient under different material regions and geometric structures. The specific calculation formula is:

[0071]

[0072] where ΔPf is the pressure drop of the fth pleat cell (Pa), μ is the fluid viscosity (Pa·s), Lf is the effective length of the fluid passing through the cell (m), Vf is the flow rate in the cell (m / s), kf is the permeability of the cell (m2), and ρ is the fluid density (kg / m3). f f f f 2 where ΔP f is the pressure drop of the fth pleat cell (Pa), μ is the fluid viscosity (Pa·s), L f is the effective length of the fluid passing through the cell (m), V f is the flow rate in the cell (m / s), kf is the permeability of the cell (m2), and ρ is the fluid density (kg / m3 ), C f is the structure-related resistance coefficient (dimensionless). By solving this system of equations, the pressure drop values and the corresponding airflow flux (Q f ) distribution for each pleat cell are calculated. The relationship between airflow flux and pressure drop and permeability is:

[0073]

[0074] where Q f is the airflow volume flux through the fth pleat cell (m 3 / s), A f is the cross-sectional area of the cell (m 2 ).

[0075] Based on the airflow flux distribution, identify the pressure imbalance areas, adjust the pleat angle parameters by gradient, and generate the angle optimization data table. Pressure imbalance areas refer to parts of the filter bag where airflow flux distribution is uneven, usually manifested as some pleat cells having too much airflow flux while others have too little. The identification process first calculates the average airflow flux (\bar{Q}) of the entire filter bag, then calculates the airflow flux deviation of each cell:

[0076]

[0077] When δ f exceeds the set threshold (usually 15%), the cell is marked as a pressure imbalance area. For these areas, the gradient descent algorithm is used to adjust the pleat angle parameters. The adjustment process follows the following formula:

[0078]

[0079] where θ new and θ old are the adjusted and unadjusted pleat angles (degrees), η is the learning rate (a parameter that controls the adjustment step size), is the gradient of the airflow uniformity objective function with respect to the angle. The objective function is usually defined as the sum of the squares of the airflow flux deviations of each cell:

[0080]

[0081] where N is the total number of pleat cells. Through multiple iterations of calculation, the pleat angle is gradually optimized until the airflow flux distribution reaches the desired uniformity. After each iteration, the angle parameters and corresponding airflow uniformity indicators are recorded in the angle optimization data table.

[0082] The angle optimization data table is combined with the wrinkle depth variable to determine the depth-density relationship curve through response surface analysis to form a wrinkle geometric coupling relationship graph. Response surface analysis is a statistical method for studying how multiple independent variables affect dependent variables. By constructing a mathematical model, the complex relationship between the two can be described. In the design of filter bags, the wrinkle angle and depth are taken as independent variables, and the filtration efficiency and pressure drop are taken as dependent variables. By designing a series of experimental points with different combinations of angle and depth, the response values of each point are obtained. Then, a polynomial regression method is used to fit the response surface equation to analyze how the wrinkle depth affects the optimal wrinkle density. The relationship between depth and density usually exhibits nonlinear characteristics, and response surface analysis can identify the wrinkle density values that achieve optimal filtration performance at different depths. The wrinkle geometric coupling relationship graph is a graphical representation of the interdependence between wrinkle geometric parameters. The horizontal axis represents the wrinkle depth, and the vertical axis represents the optimal wrinkle density. Each point on the curve corresponds to a set of optimized wrinkle geometric parameters. Based on the wrinkle geometric coupling relationship graph, airflow uniformity evaluation and parameter iterative optimization are performed to generate self-balancing wrinkle structure parameter sets. Airflow uniformity evaluation is a quantitative analysis of the uniformity of airflow distribution in filter bags with given geometric parameters. The coefficient of variation (standard deviation of airflow flux divided by the average value) is often used as an evaluation index. Parameter iterative optimization is a cyclic optimization process that continuously adjusts the wrinkle angle, depth, and density parameters to find the optimal combination of parameters that achieve the best airflow uniformity and filtration efficiency. The specific optimization process uses genetic algorithms or particle swarm optimization algorithms, with the fitness function set as the weighted sum of airflow uniformity and filtration efficiency. Through multiple generations of evolution or iterative calculations, the global optimal solution is obtained. The self-balancing wrinkle structure parameter set is a collection of geometric parameters that can achieve automatic balancing of airflow distribution and maximize filtration efficiency in filter bags under working conditions. It includes key parameters such as optimal wrinkle angle, depth, and density, providing direct guidance for subsequent filter bag structure design and manufacturing.

[0083] In a specific embodiment, the process of performing step S104 can specifically include the following steps:

[0084] (1) Quantitatively measure the interfacial force between the candidate filter materials through a material compatibility analyzer to obtain material interface bonding strength data;

[0085] (2) Evaluate the complementarity of the capture ability of the candidate filter materials, calculate the material combination capture spectrum based on the pollutant data in the pollutant characteristic digital fingerprint library, and form a material complementarity score table;

[0086] (3) Exhaustively analyze the arrangement and combination schemes of different material layers according to the material complementarity score table, and generate a material sorting scheme library through directional filtration coefficient calculation;

[0087] (4) Combine the material sequencing scheme library with the self-balanced fold structure parameter set to determine the optimal thickness ratio of each layer of material through interlayer pressure drop coordination calculation, and construct a thickness matching relationship matrix;

[0088] (5) Cross-analyze the thickness matching relationship matrix and the material interface bonding strength data to identify the interface stress concentration point, adjust the material interlayer contact mode, and generate an interface optimization strategy table;

[0089] (6) Integrate the interface optimization strategy table with the material sequencing scheme library and the thickness matching relationship matrix to form a multi-layer composite filter structure configuration table.

[0090] Specifically, the interface force between the candidate filter materials is quantitatively measured by a material compatibility analyzer to obtain material interface bonding strength data. The material compatibility analyzer is a precision instrument specially used to measure the interface bonding characteristics between different materials. It comprehensively evaluates the material interface bonding strength through tensile testing, shear testing, and peeling testing. During the measurement process, the two materials to be tested are subjected to hot pressing or bonding treatment according to the preset conditions (such as temperature, pressure, and time) to form a composite interface, and then the bonding strength is tested in a standard test environment. Tensile testing applies tension perpendicular to the interface direction to measure the maximum stress required for interface separation; shear testing applies force parallel to the interface direction to measure the maximum shear stress required for interface sliding; and peeling testing simulates the gradual separation process in actual use to measure the energy required for unit width interface separation. The test results are integrated to form the material interface bonding strength data, which includes the interface strength values, interface failure modes, and durability assessments of different material combinations under various bonding conditions. The complementarity of the capture ability of the candidate filter materials is evaluated, and the material combination capture spectrum is calculated based on the pollutant data in the pollutant characteristic digital fingerprint library to form a material complementarity score table. The complementarity of the capture ability is the process of analyzing the capture mechanism and efficiency differences of different filter materials for various pollutants to find complementary combinations. The pollutant characteristic digital fingerprint library contains multi-dimensional information such as particle size distribution, chemical properties, and adsorption behavior of pollutants. By comparing these data with the capture characteristics of various filter materials, the capture efficiency of the material for a specific pollutant is calculated. The calculation formula of the material combination capture spectrum is:

[0091] CE ab,p = 1 - (1 - CE a,p ) · (1 - CE b,p )

[0092] Where CE ab,p represents the capture efficiency of the combination of material a and material b for pollutant p, CE a,p and CE b,pThe capture efficiency of pollutants p when material a and material b are used separately. For each possible material combination, the overall capture ability for all target pollutants is calculated, forming a complementarity score, calculated as:

[0093]

[0094] where CIS ab is the complementarity score of material a and material b, P is the total number of pollutant species, w p is the weight coefficient of pollutant p (determined according to its degree of harm or importance), and the third term is used to quantify the difference in capture ability of the two materials, the greater the difference, the stronger the complementarity. By comparing the complementarity scores of different material combinations, the combination with the highest score is selected, forming a material complementarity score table.

[0095] Based on the material complementarity score table, the permutation and combination scheme of different material layers is exhaustively analyzed, and a material ranking scheme library is generated through directional filtration coefficient calculation. Exhaustive analysis is the process of systematically evaluating all possible material layer arrangement orders. For each permutation and combination, its directional filtration coefficient (DFC) is calculated, which takes into account the effect of airflow direction on filtration efficiency. The calculation formula of DFC is:

[0096]

[0097] where {a1, a2,..., a L} represents a specific material arrangement order, L is the number of material layers, CE al,p is the basic capture efficiency of the lth layer material for pollutant p, is the directional adjustment factor, with a value range of 0.6-1.4, used to adjust the effect of airflow direction on capture efficiency. Different materials have different sensitivity to airflow direction, for example, electrostatic fibers may have higher efficiency when the airflow passes in the positive direction, while activated carbon adsorption materials are less affected by direction. By calculating the DFC values of all possible permutation and combination, the material ranking scheme library is sorted from high to low according to the score, forming a material ranking scheme library.

[0098] The material ranking scheme library is combined with the self-balancing pleat structure parameter set to determine the optimal thickness ratio of each layer material through interlayer pressure drop coordination calculation, and a thickness ratio relationship matrix is constructed. The interlayer pressure drop coordination calculation is based on the inherent resistance characteristics of each layer material and the principle of minimizing total pressure drop, to solve the optimal thickness distribution. The total pressure drop calculation formula is:

[0099]

[0100] where ΔP total is the total pressure drop (Pa), ΔP l is the pressure drop of the lth layer material, Rl is the resistance coefficient per unit thickness of the i-th layer material (Pa s / m 2 ), v is the airflow velocity (m / s), t l is the thickness of the i-th layer material (m). Under the total thickness constraint, the optimal thickness ratio under different working conditions is calculated by combining various pleat structure parameters (angle, depth, density), forming a thickness ratio relationship matrix. The rows of the matrix represent different combinations of pleat structure parameters, the columns represent the thickness ratio of each layer of material, and the matrix element values are the specific thickness ratios.

[0101] Cross-analysis of the thickness ratio relationship matrix and material interface bonding strength data is performed to identify stress concentration points at the interface, adjust the contact mode between material layers, and generate an interface optimization strategy table. Stress concentration point identification is based on finite element analysis to simulate the stress distribution generated when airflow passes through the multi-layer composite filter material. For each interface, the uniformity index and maximum stress value of the stress distribution are calculated and compared with the interface bonding strength data to identify the location where the stress level exceeds the bonding strength safety threshold. For the identified stress concentration points, the causes are analyzed and optimization strategies are designed, including adjusting the material contact mode (point contact, line contact, or surface contact), adding a transition layer, adjusting the hot pressing parameters, etc. The effect of each strategy is verified by secondary finite element analysis, and the best stress reduction effect is selected to form the interface optimization strategy table. The rows of the table represent different interface locations, the columns represent different optimization strategies, and the table content is the stress reduction effect and operation parameters of each strategy.

[0102] The interface optimization strategy table, material sequencing scheme library, and thickness ratio relationship matrix are integrated to form a multi-layer composite filter structure configuration table. The integration process uses a multi-criteria decision-making method, considering factors such as filtration efficiency, pressure drop, structural stability, and manufacturing feasibility. The multi-layer composite filter structure configuration table is a complete technical specification document that specifies key parameters such as material composition, hierarchical structure, thickness ratio, and interface treatment process of the filter bag.

[0103] In a specific embodiment, the process of performing step S105 can specifically include the following steps:

[0104] (1) Set up the temperature, humidity, and pollutant concentration by hierarchical combination through the environmental parameter control box to build a multi-dimensional test working condition matrix;

[0105] (2) Convert the structure parameters in the multi-layer composite filter structure configuration table to a digital twin prototype, load the environmental conditions in the multi-dimensional test working condition matrix, perform virtual wind tunnel testing, and obtain the initial filtration resistance data;

[0106] (3) Apply accelerated aging treatment to the digital twin prototype, determine the capture saturation curve through pollutant accumulation calculation, and draw the filtration efficiency decay map;

[0107] (4) Based on the filtration efficiency decay map, the hygroscopic deformation characteristics of the filter material under different humidity conditions are analyzed, and a humidity response strain map is constructed;

[0108] (5) The humidity response strain map is combined with the initial filtration resistance data to analyze the influence of high concentration pollutants on the saturation time of the material, and a life prediction data set is generated;

[0109] (6) The boundary condition screening and performance interpolation calculation are performed on the life prediction data set to form a full-condition performance curve set.

[0110] Specifically, the temperature, humidity, and pollutant concentration are set in a hierarchical combination by the environmental parameter control box to construct a multi-dimensional test working condition matrix. The environmental parameter control box is a closed test device that can accurately control environmental conditions and is used to simulate the performance of filter bags in various working environments. Hierarchical combination refers to discretizing parameters such as temperature, humidity, and pollutant concentration according to preset intervals, and then performing orthogonal combination to form a test working condition matrix. In specific implementation, temperature is usually divided into 5 levels (e.g., 15℃, 20℃, 25℃, 30℃, 35℃), humidity is divided into 4 levels (e.g., 30%, 50%, 70%, 90%), and pollutant concentration is divided into 3 levels (e.g., low, medium, high, corresponding to different numerical ranges). Through the combination of these parameters, a multi-dimensional test working condition matrix of 5x4x3=60 working condition points is constructed. Each working condition point represents a specific combination of environmental conditions for subsequent virtual testing. The construction of the multi-dimensional test working condition matrix uses the orthogonal experimental design method, which can comprehensively cover various possible working conditions and reduce unnecessary repeated testing, improving test efficiency. The structure parameters in the multi-layer composite filter structure configuration table are converted into a digital twin prototype, and the environmental conditions in the multi-dimensional test working condition matrix are loaded for virtual wind tunnel testing to obtain initial filtration resistance data. The digital twin prototype refers to a virtual model created in a computer environment that is consistent with the structure and performance of the physical filter bag. The conversion process first imports the parameters in the multi-layer composite filter structure configuration table, such as material type, layer thickness ratio, and pleat geometric parameters, into computer-aided design (CAD) software to construct a three-dimensional geometric model. Then, the physical property parameters of each material layer (such as permeability, porosity, and surface charge) are mapped onto the geometric model to form a complete digital twin prototype. Virtual wind tunnel testing is the use of computational fluid dynamics (CFD) software to simulate and analyze the flow and filtration performance of the digital twin prototype under different environmental conditions. During testing, the environmental conditions in the multi-dimensional test working condition matrix are input as boundary conditions into the CFD model, the inlet flow rate is set to the standard test flow rate (usually 0.1 m / s), and the outlet is set to atmospheric pressure. Then, the Navier-Stokes equation set is solved to calculate the flow field distribution and pressure field distribution. The initial filtration resistance data refers to the pressure drop of the filter bag in the early stage of use (before the influence of accumulated pollutant particles), which is obtained by calculating the average pressure difference between the inlet and outlet. Virtual wind tunnel testing is performed on the 60 working condition points to obtain an initial filtration resistance data matrix, with each element in the matrix corresponding to the initial pressure drop value under a specific environmental condition.

[0111] An accelerated aging process is applied to the digital twin prototype to determine the capture saturation curve and draw the filtration efficiency decay map. The accelerated aging process is a computational method that simulates the impact of pollutant accumulation on performance over the long-term use of the filter bag. By increasing the virtual time step and pollutant concentration, the simulation period is shortened. During the process, based on the previous CFD model, a particle trajectory tracking and deposition simulation module is added to calculate the capture process and accumulation of particles in the filter material. The pollutant accumulation calculation uses the Lagrangian-Eulerian hybrid method, which first calculates the flow field (Eulerian method) and then tracks the motion trajectories of a large number of representative particles (Lagrangian method) in this flow field, recording the position and number of captured particles. This process is carried out in sequence according to the virtual time step, and the permeability and porosity of the material are updated after each time step to reflect the changes in material performance caused by pollutant accumulation. The capture saturation curve is a curve that describes the relationship between the mass of pollutants accumulated over time and the change in capture efficiency of the filter material. The horizontal axis represents the accumulated pollutant mass (or equivalent service time), and the vertical axis represents the filtration efficiency. The filtration efficiency decay map is a collection of capture saturation curves drawn under different environmental conditions, which visually demonstrates how environmental factors affect the service life and performance decay rate of the filter bag.

[0112] Based on the filtration efficiency decay map, the hygroscopic deformation characteristics of the filter material under different humidity conditions are analyzed, and a humidity response strain map is constructed. Hygroscopic deformation characteristics refer to the properties of filter materials that change shape (such as expansion, contraction, distortion, etc.) after absorbing moisture from the air. This deformation affects the structural stability and filtration performance of the filter material. The analysis process first extracts performance data under different humidity conditions from the filtration efficiency decay map, and then combines material hygroscopic performance parameters (such as hygroscopic coefficient, saturated moisture content, etc.) to calculate the expansion rate or contraction rate of the material under each humidity condition. The humidity response strain map is a graph that describes the relationship between the strain (deformation ratio) of the filter material and the environmental humidity. It is usually drawn as multiple curves, each representing the humidity-strain relationship at a specific temperature. The construction process uses multivariate regression analysis to consider the effects of temperature, humidity, and pollutant accumulation on material deformation, establishes a mathematical model, and draws the corresponding curve graph. This graph has important reference value for predicting the structural changes and performance fluctuations of the filter bag in high humidity environments.

[0113] The humidity response strain map is combined with the initial filtration resistance data to analyze the impact of high concentration pollutants on the saturation time of the material, generating a life prediction dataset. The deformation of the filtration material caused by humidity changes directly affects its filtration resistance and capture efficiency, combined with the rapid accumulation of high concentration pollutants, both of which determine the service life of the filter bag. The analysis process first establishes a comprehensive evaluation model to convert the deformation data in the humidity response strain map into a resistance change factor, which is multiplied by the initial filtration resistance data to obtain the actual resistance value considering the influence of humidity. Then, combined with the pollutant accumulation rate in the capture saturation curve (proportional to concentration), the time required for the filtration material to reach saturation state (usually defined as the filtration efficiency dropping to 80% of the initial value or the pressure drop increasing to twice the initial value) is calculated, i.e. the saturation time of the material. For each working condition point in the multi-dimensional test working condition matrix, the corresponding saturation time is calculated to form the life prediction dataset. This dataset is a multi-dimensional array that records the expected service life of the filter bag under different combinations of temperature, humidity, and pollutant concentration.

[0114] Boundary condition screening and performance interpolation calculation are performed on the life prediction dataset to form a full-condition performance curve set. Boundary condition screening is to select representative extreme condition points from the multi-dimensional test working condition matrix, which are usually located at the boundaries or turning points of the parameter space, reflecting the main characteristics of the performance curve. The screening method combines extreme value analysis and sensitivity analysis to find the working condition points that have the most significant impact on the service life of the filter bag. Performance interpolation calculation is based on the known boundary condition point data to calculate the performance parameters of intermediate condition points. The interpolation process uses multi-dimensional spline interpolation or Kriging interpolation method, which can accurately reflect the nonlinear impact of parameter changes on performance while ensuring data smoothness. The full-condition performance curve set is a set of performance curves covering all possible working conditions, including filtration efficiency curves, pressure drop change curves, and life prediction curves. These curves present the performance characteristics and service life of the filter bag under various environmental conditions in an intuitive graphical manner, providing scientific basis for the selection and application of filter bags.

[0115] For example, when designing a filter bag suitable for high humidity industrial environments, first set 5 temperature levels (15℃, 20℃, 25℃, 30℃, 35℃), 4 humidity levels (30%, 50%, 70%, 90%) and 3 pollutant concentration levels (100μg / m 3 , 500μg / m 3 , 1000μg / m 3), a multi-dimensional test condition matrix containing 60 working points was constructed. Then the optimal design scheme (three-layer structure of electrostatic fiber + activated carbon fiber + HEPA fiber, thickness ratio of 2:3:1, pleat angle of 22°, depth of 20 mm, and density of 4 pleats / cm) was extracted from the multi-layer composite filter structure configuration table and converted into a digital twin prototype. During the virtual wind tunnel test, the standard flow rate of 0.1 m / s was selected, and the initial filtration resistance was calculated for 60 working points to obtain the initial pressure drop matrix. The results show that, at 25°C, the initial pressure drop increases by about 25% when the humidity increases from 30% to 90%, mainly due to the decrease in porosity caused by the swelling of the material due to moisture absorption. Then, the digital twin prototype was subjected to accelerated aging treatment to simulate a 6-month use process and track the accumulation of pollutant particles. Through calculation, under the condition of high pollutant concentration (1000 μg / m 3 ), the capture efficiency of the filter material decreases from 99% to 85% after 1000 hours, while under the condition of low concentration (100 μg / m 3 ), the same efficiency decay requires about 10000 hours, showing an approximately linear proportional relationship. Based on the filtration efficiency decay map, the moisture absorption and deformation characteristics of the filter material were analyzed, and it was found that when the humidity exceeds 70%, the volume expansion rate of the electrostatic fiber layer reaches 8%, leading to a decrease in porosity and an increase in pressure drop, which in turn affects the capture efficiency. The humidity response strain map shows that the higher the temperature, the more sensitive the material is to humidity changes. Under the condition of 35°C and 90% humidity, the comprehensive strain rate is the largest. By combining the humidity response strain map with the initial filtration resistance data, the saturation time of the filter bag under different working conditions was calculated, and a life prediction data set was generated. The data shows that under the most severe conditions (35°C, 90% humidity, 1000 μg / m 3 pollutant concentration), the service life of the filter bag is about 800 hours, while under the optimal conditions (15°C, 30% humidity, 100 μg / m 3 pollutant concentration), the life can be extended to 12000 hours. By selecting 8 representative working points from the life prediction data set and using the cubic spline interpolation method to calculate the performance parameters of the intermediate working points, a full working condition performance curve set covering all possible working conditions was finally formed, providing accurate performance reference and life prediction for the actual application of the filter bag.

[0116] In a specific embodiment, the process of performing step S106 can specifically include the following steps:

[0117] (1) Reverse analysis of material performance indicators in the full working condition performance curve set by a material formula parser to determine the material composition ratio relationship and generate a material formula list;

[0118] (2) Perform availability assessment on the formulation items in the material formulation list, match the existing supplier resource library, and form a material procurement parameter table;

[0119] (3) Based on the correspondence between the multi-layer composite filter structure configuration table and the full working condition performance curve set, extract key structure size data, and construct a manufacturing size specification card;

[0120] (4) Convert the manufacturing size specification card into production equipment parameters, determine the equipment adjustment parameters through process adaptability analysis, and form a process flow instruction set;

[0121] (5) Use the process flow instruction set to carry out small batch trial production of filter bag samples, collect actual production deviation data, and establish a process correction factor table;

[0122] (6) Based on the process correction factor table, fine-tune and verify the material formulation list and manufacturing size specification card, and integrate to form a filter bag design scheme that can be directly used for production.

[0123] Specifically, the material formulation parser is used to perform reverse analysis on the material performance indicators in the full working condition performance curve set, determine the material composition ratio relationship, and generate a material formulation list. The material formulation parser is a data analysis tool based on machine learning algorithm, which is specially used to deduce the composition ratio of filter material from its performance characteristics. The reverse analysis process uses a multi-objective optimization algorithm, taking performance indicators such as filter efficiency, pressure drop, and service life in the full working condition performance curve set as objective functions, and taking material composition ratio as optimization variables, to establish a performance-component mapping relationship model. In specific implementation, first, extract the component-performance correspondence of known materials from the material database as training samples, and use machine learning methods such as support vector regression or random forest to establish a prediction model; then use genetic algorithm or particle swarm optimization method to search in the component space, find the material ratio combination that best matches the target performance indicators. The generated material formulation list contains detailed information of various filter material layers such as fiber type, additive type and content, surface treatment agent formulation, etc., providing accurate guidance for material production.

[0124] The material procurement parameter table is formed by performing availability evaluation on the formulation items in the material formulation list, matching the existing supplier resource library. The availability evaluation is a process of analyzing the supply status of each component in the material formulation on the market, aiming to ensure that the designed filter material can be actually produced. The evaluation process first matches each component in the formulation list with the supplier resource library, which is a structured database containing information of material suppliers, recording the supplier name, material specification, price, minimum order quantity, delivery cycle, etc. Matching uses fuzzy query and similarity calculation method to find out the supplier list that can provide similar or equivalent materials. For the components that match successfully, the supplier information is directly recorded; for the components that fail to match, alternative scheme analysis is performed to find alternative materials with similar performance but easier to obtain. The final material procurement parameter table contains detailed procurement specifications, recommended suppliers, price range, ordering cycle, etc. of each material, facilitating the production department to efficiently and accurately purchase raw materials.

[0125] Based on the correspondence between the multi-layer composite filter structure configuration table and the full- working condition performance curve set, key structure size data is extracted, and a manufacturing size specification card is constructed. Key structure size data refers to geometric parameters that have a significant impact on filter bag performance, including the thickness of each material layer, the pleat angle, the pleat depth, the pleat density, the overall size, etc. The extraction process first analyzes the optimal design scheme in the multi-layer composite filter structure configuration table, and then combines the performance data in the full- working condition performance curve set to determine the tolerance range of each parameter. Tolerance calculation is based on the sensitivity analysis method, which quantifies the influence of each parameter change on performance by calculating the partial derivative of performance with respect to each parameter, and then determines the reasonable tolerance value. The manufacturing size specification card is a technical document that records the size requirements of each part of the product in detail, including standard size, tolerance range, detection method, etc. The construction of the specification card adopts hierarchical structure design, which expands from overall size to local details layer by layer, ensuring the size precision control in the manufacturing process. The manufacturing size specification card is converted into production equipment parameters, and the equipment adjustment parameters are determined through process adaptability analysis to form a process flow instruction set. Process adaptability analysis is a process to evaluate whether a specific design can be implemented on existing equipment, with the goal of identifying potential production difficulties and making adjustments in advance. The analysis process first compares the size requirements in the manufacturing size specification card with the processing capacity of the production equipment, including machining precision, speed range, temperature control range, pressure control range, etc. The comparison uses the difference analysis method to calculate the gap between design requirements and equipment capacity, and to determine whether the equipment parameters need to be adjusted or the design requirements need to be modified. For equipment parameters that need to be adjusted, the best setting value is calculated to meet the design requirements, forming a list of equipment adjustment parameters. The process flow instruction set is a technical document that contains complete production steps and equipment parameter settings, which specifies the operation method, equipment parameter and quality control point of each link from raw material handling to finished product inspection in detail, ensuring the consistency of the production process and the stability of the product quality.

[0126] A small batch trial is conducted on the filter bag sample using the process flow instruction set to collect actual production deviation data and establish a process correction factor table. The small batch trial is a process of verifying the design scheme before formal production, usually producing 10-20 sample products for testing the feasibility of the design and the stability of the process. The trial process is strictly carried out according to the process flow instruction set, while recording the actual parameters and intermediate product characteristics at each link. The actual production deviation data refers to the difference between the trial product and the design target, including size deviation, performance deviation, appearance deviation, etc. The data collection adopts a multi-point detection method to measure the key positions of each sample in detail to form a complete deviation data set. The process correction factor is a parameter used to compensate for systematic deviations in the production process, which is calculated by statistical analysis of the actual production deviation data. The calculation process uses regression analysis method to establish the mathematical relationship between the deviation and the process parameters, and to deduce the correction factor that can minimize the deviation. The process correction factor table records the correction parameters and applicable conditions of each process link, providing accurate parameter adjustment basis for subsequent mass production.

[0127] Based on the process correction factor table, the material formula list and the manufacturing size specification card are fine-tuned and verified, and are integrated to form a filter bag design scheme that can be directly used for production. Fine-tuning and verification is a process of final optimization of the design scheme based on the small batch trial results, to ensure the feasibility and stability of the design scheme in the actual production environment. Fine-tuning of the material formula mainly considers the difference between the actual performance and the expected performance of the material, and adjusts the proportion of ingredients or adds auxiliary materials to make up for the deficiency; the verification of the manufacturing size is based on the size deviation data, and uses the compensation design method, that is, the influence of production deviation is considered in advance in the design size to obtain more accurate final size. The integration process unifies the corrected material formula list, manufacturing size specification card and process flow instruction set to form a complete design scheme document. The filter bag design scheme that can be directly used for production is a comprehensive technical document, which includes four main parts: product design description, material procurement specification, manufacturing process guide and quality control standard.

[0128] The above describes the filter bag intelligent design method based on artificial intelligence in the embodiments of the present application, and the following describes the filter bag intelligent design system based on artificial intelligence in the embodiments of the present application, please refer to Figure 2 An embodiment of the filter bag intelligent design system based on artificial intelligence in the embodiments of the present application includes:

[0129] The acquisition module is configured to acquire multi-dimensional data of particle size distribution and adsorption characteristics of indoor air pollutants by using the pollutant particle dynamic capture device, and obtain a pollutant characteristic digital fingerprint library.

[0130] a quantification module for quantifying the porosity, active site distribution, and surface charge of the filter material using a material microscopic permeability measuring instrument based on the pollutant characteristic digital fingerprint library, to obtain a multi-dimensional material performance matrix;

[0131] a calculation module for parameterizing modeling and airflow distribution calculation of the filter bag pleat angle, depth, and density based on the multi-dimensional material performance matrix, to obtain a self-balancing pleat structure parameter set;

[0132] an adjustment module for optimizing and adjusting the combination order, thickness ratio, and contact interface of different material layers of the filter bag based on the self-balancing pleat structure parameter set, to obtain a multi-layer composite filter structure configuration table;

[0133] a test module for simulating the filtration efficiency and service life of the filter bag in a high-humidity and high-concentration pollutant environment using the multi-layer composite filter structure configuration table, to obtain a full-working-condition performance curve set;

[0134] a verification module for accurately converting and verifying the filter bag material formula, structure size, and production process parameters based on the full-working-condition performance curve set, to obtain a filter bag design scheme that can be directly used for production.

[0135] Through the collaborative efforts of the aforementioned components, a multi-dimensional data collection system for indoor air pollutants was implemented using a dynamic pollutant particle capture device. This system established a precise digital fingerprint database of pollutant characteristics, containing key information such as particle size distribution and adsorption properties. This database provides an accurate data foundation for subsequent design, addressing the problem of insufficient understanding of pollutant characteristics in traditional design. Secondly, a material micro-permeability measuring instrument was used to precisely quantify the filter material, generating a multi-dimensional material performance matrix. This matrix comprehensively characterizes the material's porosity, active site distribution, and surface charge, enabling designers to more accurately select filter materials suitable for specific pollutants. Furthermore, parametric modeling and airflow distribution calculations based on the multi-dimensional material performance matrix generated a self-balancing pleated structure parameter set. Artificial intelligence algorithms were used to optimize parameters such as pleat angle, depth, and density, achieving optimal filter bag structure design and significantly improving filtration efficiency and lifespan. In addition, the combination sequence, thickness ratio, and contact interface of different material layers in the filter bag were optimized and adjusted to form a multi-layer composite filter structure configuration table. This fully leveraged the synergistic effect of each material layer, enhancing the overall performance of the filtration system. Of particular note is the application of artificial intelligence algorithms in this solution for virtual testing and lifespan prediction. By simulating the performance of filter bags in high-humidity, high-concentration pollutant environments, a full-condition performance curve set is generated, accurately predicting the product's performance and lifespan under various environmental conditions, avoiding the limitations of traditional methods that rely on physical testing. Finally, based on the precise conversion and verification of the full-condition performance curve set, a filter bag design scheme that can be directly used for production is generated, achieving a seamless connection from design to production and improving design conversion efficiency and product consistency. The application of artificial intelligence algorithms in this invention, especially the deep learning and optimization algorithms in pollutant data fusion processing, material performance prediction, structural parameter optimization, and lifespan prediction, greatly enhances the intelligence and accuracy of the design process. This allows filter bag design to move beyond experience and trial and error, relying instead on precise data and scientific calculations to achieve personalized and precise design for complex indoor air pollution environments, thereby significantly improving the filtration efficiency, lifespan, and environmental adaptability of filter bags.

[0136] Reference Figure 3 This invention also provides a computer device, which can be a server, and its internal structure can be as follows: Figure 3As shown, the computer device includes a processor, memory, display screen, input device, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database stores the data corresponding to this embodiment. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements the above-described method.

[0137] Those skilled in the art will understand that Figure 3 The structures shown are merely block diagrams of some structures related to the present invention and do not constitute a limitation on the computer devices on which the present invention is applied.

[0138] An embodiment of the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method. It is understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0139] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the present invention and embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM, etc.

[0140] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0141] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0142] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for intelligent design of filter bags based on artificial intelligence, characterized in that, The AI-based intelligent design method for filter bags includes: The particle size distribution and adsorption characteristics of indoor air pollutants are collected in multiple dimensions by a dynamic pollutant particle capture device to obtain a digital fingerprint database of pollutant characteristics. Based on the pollutant characteristic digital fingerprint database, the porosity, active site distribution, and surface charge of the filter material are precisely quantified using a material micro-permeability measuring instrument to obtain a multi-dimensional material performance matrix; Based on the multidimensional material property matrix, the filter bag pleat angle, depth, and density are parametrically modeled and the airflow distribution is calculated to obtain a self-balancing pleat structure parameter set. Based on the self-balancing pleated structure parameter group, the combination order, thickness ratio and contact interface of different material layers of the filter bag are optimized and adjusted to obtain a multi-layer composite filter structure configuration table. The filtration efficiency and service life of the filter bag under high humidity and high concentration of pollutants were simulated and tested using the configuration table of the multi-layer composite filter structure, and a set of full-condition performance curves were obtained. Based on the full-condition performance curve set, the filter bag material formula, structural dimensions, and production process parameters are accurately converted and verified to obtain a filter bag design scheme that can be directly used for production.

2. The intelligent filter bag design method based on artificial intelligence according to claim 1, characterized in that, The process involves collecting multi-dimensional data on the particle size distribution and adsorption characteristics of indoor air pollutants using a dynamic pollutant particle capture device, resulting in a digital fingerprint database of pollutant characteristics, including: A high-precision light scattering sensor is used to monitor the concentration of PM2.5 particles in indoor air in real time and obtain a histogram of particle size distribution. A gas chromatograph was used to measure the concentration gradient of indoor formaldehyde and VOC gaseous pollutants, and a pollutant concentration change curve was generated. The surface adhesion of different pollutants under various temperature and humidity conditions was quantitatively recorded using an electrostatic adsorption test plate, and a pollutant adhesion characteristic dataset was constructed. Based on the obtained particle size distribution histogram, pollutant concentration change curve and pollutant adhesion characteristic dataset, a multi-source data correlation analysis is performed using a data fusion processor to generate a spatiotemporal distribution feature map of pollutants. The generated spatiotemporal distribution feature map of pollutants is imported into the feature extraction unit to extract key pollutant feature parameters and establish a pollutant behavior pattern library. Based on the established pollutant behavior pattern database and historical environmental condition data, cross-validation and normalization are performed to form a digital fingerprint database of pollutant characteristics.

3. The intelligent filter bag design method based on artificial intelligence according to claim 1, characterized in that, The process involves using a material microscopic permeability measuring instrument to precisely quantify the porosity, active site distribution, and surface charge of the filter material based on the pollutant characteristic digital fingerprint database, resulting in a multidimensional material performance matrix, including: The specific surface area of ​​the filter material was measured by nitrogen adsorption, and the pore size distribution curve of the material was recorded. Electron microscopy was used to perform high-magnification imaging of the surface morphology of the filter material to obtain a microstructure map of the material surface. The adsorption capacity of the filter material under different temperature conditions was quantitatively tested using a thermogravimetric analyzer, and an adsorption thermodynamic characteristic diagram was generated. The material pore size distribution curve and the material surface microstructure map are fused to establish a three-dimensional pore network topology of the material. The active sites in the three-dimensional pore network topology of the material were located and their density was measured by fluorescent labeling, and a spatial distribution map of the active sites was constructed. By combining the spatial distribution map of the active sites with the adsorption thermodynamic property map, a material capture efficiency prediction model is constructed through correlation analysis, generating a multidimensional material performance matrix.

4. The intelligent filter bag design method based on artificial intelligence according to claim 1, characterized in that, The parameter modeling and airflow distribution calculation of the filter bag pleat angle, depth, and density based on the multidimensional material property matrix yields a set of self-balancing pleat structure parameters, including: The initial geometry of the folded structure is discretized using a geometry parameter generator to form a set of folded unit grid points. The material physical parameters in the multidimensional material property matrix are mapped onto the set of folded unit grid points to construct a digital folded material model. A flow field boundary condition is applied to the digital folded material model, and the airflow distribution of each folded unit is determined by segmented pressure drop calculation. Based on the airflow distribution, regions with uneven pressure are identified, and the fold angle parameters are adjusted in a gradient to generate an angle optimization data table. By combining the angle optimization data table with the fold depth variable, the depth-density relationship curve is determined through response surface analysis, forming a fold geometric coupling relationship diagram. Based on the fold geometry coupling diagram, airflow uniformity is evaluated and parameters are iteratively optimized to generate a self-balancing fold structure parameter set.

5. The intelligent filter bag design method based on artificial intelligence according to claim 1, characterized in that, The step involves optimizing the combination order, thickness ratio, and contact interface of different material layers of the filter bag based on the self-balancing pleated structure parameter set, resulting in a multi-layer composite filter structure configuration table, including: The interfacial forces between candidate filter materials are quantitatively measured using a material compatibility analyzer to obtain data on the interfacial bonding strength. The candidate filter materials are evaluated for their capture capacity complementarity. Based on the pollutant data in the pollutant characteristic digital fingerprint database, the capture spectrum of the material combination is calculated to form a material complementarity scoring table. Based on the aforementioned material complementarity scoring table, an exhaustive analysis of the permutation and combination schemes of different material layers is performed, and a material ranking scheme library is generated by calculating the directional filtering coefficient. By combining the material sorting scheme library with the self-balancing folded structure parameter group, the optimal thickness ratio of each layer of material is determined through interlayer pressure drop coordination calculation, and a thickness ratio relationship matrix is ​​constructed. By cross-analyzing the thickness ratio matrix and the material interface strength data, stress concentration points at the interface are identified, the interlayer contact mode of the material is adjusted, and an interface optimization strategy table is generated. The interface optimization strategy table is integrated with the material sorting scheme library and the thickness ratio relationship matrix to form a multi-layer composite filter structure configuration table.

6. The intelligent filter bag design method based on artificial intelligence according to claim 1, characterized in that, The filtration efficiency and service life of the filter bag under high humidity and high concentration of pollutants were simulated and tested using the multi-layer composite filter structure configuration table, resulting in a set of full-condition performance curves, including: By using an environmental parameter control box, temperature, humidity, and pollutant concentration can be set in a graded combination to construct a multi-dimensional test condition matrix. The structural parameters in the configuration table of the multi-layer composite filter structure are converted into a digital twin prototype, the environmental conditions in the multi-dimensional test condition matrix are loaded, a virtual wind tunnel test is performed, and the initial filter resistance data is obtained. Accelerated aging treatment was applied to the digital twin prototype, and the capture saturation curve was determined by calculating the amount of accumulated pollutants, and a filtration efficiency decay spectrum was plotted. Based on the filtration efficiency decay spectrum, the moisture absorption deformation characteristics of the filter material under different humidity conditions are analyzed, and a humidity response strain diagram is constructed. By combining the humidity response strain map with the initial filter resistance data, the impact of high concentration pollutants on the material saturation time is analyzed, and a lifetime prediction dataset is generated. Boundary condition screening and performance interpolation calculations are performed on the life prediction dataset to form a full-condition performance curve set.

7. The intelligent filter bag design method based on artificial intelligence according to claim 1, characterized in that, The process involves precisely converting and verifying the filter bag material formulation, structural dimensions, and production process parameters based on the full-condition performance curve set, resulting in a filter bag design scheme that can be directly used for production, including: The material performance indicators in the full-condition performance curve set are analyzed in reverse by the material formula parser to determine the material composition ratio and generate a material formula list. Availability assessments are performed on the formulation items in the material formulation list, and existing supplier resource pools are matched to form a material procurement parameter table. Based on the correspondence between the configuration table of the multi-layer composite filter structure and the set of full-condition performance curves, key structural dimension data are extracted to construct a manufacturing dimension specification card. The manufacturing dimension specification card is converted into production equipment parameters, and the equipment adjustment parameters are determined through process adaptability analysis to form a process flow instruction set. The filter bag samples were produced in small batches using the process flow instruction set. Actual production deviation data were collected, and a process correction factor table was established. Based on the process correction factor table, the material formula list and the manufacturing size specification card are fine-tuned and verified to form a filter bag design scheme that can be directly used for production.

8. An AI-based intelligent filter bag design system, used to implement the AI-based intelligent filter bag design method as described in any one of claims 1-7, characterized in that, The AI-based intelligent filter bag design system includes: The data acquisition module is used to collect multi-dimensional data on the particle size distribution and adsorption characteristics of indoor air pollutants through a dynamic pollutant particle capture device, and obtain a digital fingerprint database of pollutant characteristics. The quantification module is used to accurately quantify the porosity, active site distribution, and surface charge of the filter material using a material micro-permeability measuring instrument based on the pollutant characteristic digital fingerprint database, thereby obtaining a multi-dimensional material performance matrix; The calculation module is used to perform parametric modeling and airflow distribution calculation on the pleat angle, depth and density of the filter bag based on the multidimensional material property matrix, and to obtain a self-balancing pleat structure parameter set. The adjustment module is used to optimize and adjust the combination order, thickness ratio, and contact interface of different material layers of the filter bag according to the self-balancing pleated structure parameter group, so as to obtain a multi-layer composite filter structure configuration table. The testing module is used to simulate and test the filtration efficiency and service life of the filter bag under high humidity and high concentration of pollutants using the configuration table of the multi-layer composite filter structure, and obtain a set of full-condition performance curves. The verification module is used to accurately convert and verify the filter bag material formula, structural dimensions, and production process parameters based on the full-condition performance curve set, so as to obtain a filter bag design scheme that can be directly used for production.

9. A computer device, characterized in that, The device includes a memory and a processor, the memory storing a computer program that can run on the processor, characterized in that the processor, when executing the computer program, implements the intelligent design method for filter bags based on artificial intelligence as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, the computer program causing a processor, when executed by a processor, to perform the intelligent design method for filter bags based on artificial intelligence as described in any one of claims 1 to 7.

Images