A rapid detection method for zeolite loading and uniformity in zeolite composite material

Abstract

The application discloses a kind of zeolite loadings and uniformity of zeolite composite material fast detection method, specifically related to material performance detection technical field;The sample of the composite material to be measured is fluorescently labeled and processed;Obtain fluorescence image under specific excitation wavelength;Three-dimensional light intensity distribution atlas is constructed by image processing algorithm;Nonlinear fitting relationship between fluorescence intensity and standard zeolite loading is established by using local weighted regression model;Further estimate the overall zeolite loading of sample;At the same time, based on fluorescence integration, distribution skewness and anisotropy analysis, the zeolite distribution coefficient of variation is calculated, and the uniformity of its distribution is judged accordingly;The detection method of the application has the advantages of rapidity, high sensitivity, non-destructive, etc., and can realize automatic image recognition and quantitative evaluation, and is suitable for zeolite distribution detection of various types of composite materials, and has wide application prospect in the preparation and quality control of catalysts, adsorbents, zeolite membranes and other materials.

Description

Technical Field

[0001] This invention relates to the field of material performance testing technology, specifically to a rapid method for detecting the zeolite loading and uniformity in zeolite composite materials. Background Technology

[0002] Zeolites are a class of crystalline aluminosilicate materials with regular porous structures, widely used in catalysts, adsorbents, ion exchangers, and other fields. To meet the performance requirements of specific industrial applications, zeolites are often loaded onto various composite matrices to form zeolite composites. The loading amount and uniformity of zeolite distribution in the composite material have a significant impact on the final material's properties.

[0003] Currently, the main methods for detecting the zeolite loading and its uniformity in zeolite composites include thermogravimetric analysis (TGA), scanning electron microscopy (SEM) combined with energy dispersive spectroscopy (EDS), and X-ray diffraction (XRD). However, these methods generally suffer from problems such as complex operation, long detection cycle, high sample pretreatment requirements, and high cost, making it difficult to meet the needs of rapid, large-scale detection.

[0004] Therefore, there is an urgent need for a simple, rapid, and accurate method to quantitatively or semi-quantitatively detect the zeolite loading and its distribution uniformity in zeolite composites, so as to improve the efficiency of quality control in the production process and provide a reliable basis for the evaluation of material properties. Summary of the Invention

[0005] The purpose of this invention is to provide a rapid detection method for zeolite loading and uniformity in zeolite composite materials, so as to overcome the shortcomings of the prior art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a rapid detection method for zeolite loading and uniformity in zeolite composite materials, comprising:

[0007] The zeolite composite material sample to be tested is brought into contact with a solution containing fluorescent ligands to form a fluorescent label;

[0008] Fluorescence imaging of the sample is performed under selected wavelength excitation to obtain an optical image of the marked area;

[0009] Using image processing algorithms, the acquired optical images are rasterized and a three-dimensional light intensity distribution map is constructed;

[0010] Based on the three-dimensional light intensity distribution map, the coverage of zeolite and the mean deviation of light intensity in each region are quantified, and the zeolite distribution variation coefficient is output to evaluate the uniformity of zeolite distribution in the composite material.

[0011] A fitting curve was established between fluorescence signal intensity and standard zeolite loading. Based on the fitting curve, the image grayscale value was correlated with the actual zeolite mass content, and the zeolite loading was estimated.

[0012] Preferably, the image processing algorithm includes:

[0013] The optical fluorescence image obtained by excitation is processed into a two-dimensional pixel matrix, and the image area is divided into grids according to a preset spatial resolution to construct a light intensity data raster unit.

[0014] Local variance normalization is performed on the light intensity value of each pixel in each grid cell;

[0015] A convolutional nested spline interpolation model is used to perform longitudinal interpolation fitting on the light intensity data of each grid cell to generate a continuous three-dimensional light intensity distribution map.

[0016] Preferably, the calculation of the zeolite distribution variation coefficient includes:

[0017] Based on the three-dimensional light intensity distribution map, the map is divided into multiple sub-volume regions according to spatial hierarchy, and the fluorescence intensity integral value of zeolite in each sub-region is extracted.

[0018] Construct a probability density distribution model of zeolite light intensity in the overall space, and calculate its distribution skewness and kurtosis;

[0019] By introducing a spherical scanning window structure, the local mean change rate is statistically analyzed by moving the window in each direction, thus forming an anisotropic distribution index.

[0020] The coefficient of variation of zeolite distribution is output through a multi-factor weighted model by integrating the fluorescence intensity integral value, distribution skewness and kurtosis, and anisotropic distribution index, and is used as the result of zeolite uniformity detection.

[0021] Preferably, the obtained zeolite distribution variation coefficient is compared with a preset threshold. If the zeolite distribution variation coefficient is greater than the preset threshold, it indicates that the zeolite is not uniformly distributed in the composite material; if the zeolite distribution variation coefficient is less than or equal to the preset threshold, it indicates that the zeolite is uniformly distributed in the composite material.

[0022] Preferably, the establishment of the fitting curve between the fluorescence signal intensity and the standard zeolite loading includes:

[0023] Multiple sets of composite material standard samples with known zeolite mass fractions were selected, and their corresponding average fluorescence intensity values ​​were obtained through the same fluorescent labeling and image acquisition process;

[0024] Outlier removal and light intensity signal normalization were performed on the collected data;

[0025] Based on a local weighted regression model, the nonlinear mapping relationship between fluorescence intensity and actual zeolite loading is fitted.

[0026] Model parameters were optimized using cross-validation and the minimum mean square error criterion.

[0027] Preferably, the establishment of the nonlinear mapping relationship is based on a locally weighted regression model, specifically including:

[0028] For each standard sample point, a local neighborhood sample subset is dynamically selected based on the position of its fluorescence intensity value in the input space;

[0029] In each local region, a first-order or second-order polynomial regression model is independently fitted to obtain the estimated values ​​of fluorescence intensity and zeolite loading in that region.

[0030] Multiple local regression results are combined to form a global fitting curve.

[0031] Preferably, estimating the zeolite loading includes:

[0032] The three-dimensional light intensity distribution map obtained through image processing is channel normalized, and the average gray value of each grid voxel region is extracted;

[0033] The average gray value is used as an input variable and substituted into the fitting relationship curve constructed based on the local weighted regression model to obtain the corresponding predicted value of zeolite mass fraction.

[0034] The predicted values ​​for all voxel regions are volume-weighted averaged to calculate the total zeolite loading of the entire sample.

[0035] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0036] 1. This invention, by introducing specific fluorescent labeling and high-resolution image processing technology, combined with nonlinear modeling and spatial distribution analysis, enables the visualization, quantification, and high-throughput identification of the loading state of zeolites in composite structures. Compared with existing thermogravimetric analysis, electron microscopy, or X-ray analysis methods, this invention is simple to operate, has a short detection cycle, and provides rich data structures, making it suitable for online rapid screening and industrial process quality control.

[0037] 2. This invention quantifies the spatial distribution characteristics of zeolites from multiple dimensions (integral light intensity, distribution skewness, and anisotropy) by constructing a three-dimensional fluorescence spectrum and a coefficient of variation model. Furthermore, it employs a local weighted regression method to establish a nonlinear mapping relationship between fluorescence signals and loading, significantly improving prediction accuracy and model generalization ability. This method does not rely on expensive instruments, possesses broad applicability and strong engineering practicality, and can be widely applied to structural analysis and homogeneity assessment in various fields such as catalytic materials, adsorbents, and composite membranes. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0039] Figure 1 This is a mind map of the method of the present invention. Detailed Implementation

[0040] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0041] Example 1, please refer to Figure 1 As shown in this embodiment, a rapid detection method for zeolite loading and uniformity in a zeolite composite material includes:

[0042] The zeolite composite material sample to be tested is brought into contact with a solution containing fluorescent ligands to form a fluorescent label;

[0043] Fluorescence imaging of the sample is performed under selected wavelength excitation to obtain an optical image of the marked area;

[0044] Using image processing algorithms, the acquired optical images are rasterized and a three-dimensional light intensity distribution map is constructed;

[0045] Based on the three-dimensional light intensity distribution map, the coverage of zeolite and the mean deviation of light intensity in each region are quantified, and the zeolite distribution variation coefficient is output to evaluate the uniformity of zeolite distribution in the composite material.

[0046] A fitting curve was established between fluorescence signal intensity and standard zeolite loading. Based on the fitting curve, the image grayscale value was correlated with the actual zeolite mass content, and the zeolite loading was estimated.

[0047] This embodiment provides a method for fluorescently labeling zeolite composite material samples, which is used to subsequently detect the loading and distribution of zeolite in the composite material using optical means.

[0048] Zeolite composite material samples to be tested (such as alumina-based particles loaded with ZSM-5).

[0049] Fluorescent ligand solution: 0.1 mM of 4-carboxyfluorescein dissolved in a mixture of ethanol and water (volume ratio 7:3);

[0050] Buffer solution: 50 mM phosphate buffer, pH = 6.8;

[0051] Standard laboratory equipment includes glass dishes, magnetic stirrers, constant temperature water baths, centrifuges, and drying ovens.

[0052] Sample pretreatment: 5 g of zeolite composite material sample was dried in a 60 °C oven for 4 hours to remove physically adsorbed moisture;

[0053] Fluorescent ligand reaction: Take the dried sample and add it to 20 mL of fluorescent ligand solution. Stir magnetically for 2 hours at room temperature to allow the carboxylic acid groups in the fluorescent ligand to undergo esterification or hydrogen bonding with the acidic hydroxyl groups (Si–OH / Al–OH) on the zeolite surface to form a stable surface label.

[0054] Elution and drying: After centrifugation, the sample was washed three times with ethanol to remove unbound fluorescent molecules, and then vacuum dried at 40 °C for 6 hours to obtain the fluorescently labeled zeolite composite material.

[0055] Under weakly acidic conditions, the carboxyl group in the fluorescent ligand can form a hydrogen bond complex with the hydroxyl group on the zeolite surface or form an ester bond through condensation. This reaction process can be expressed as: zeolite–OH + R–COOH → zeolite–OOC–R + H₂O; where R represents the fluorescein molecule group. The labeled zeolite emits green light (approximately 520 nm) when excited at a wavelength of 490 nm, facilitating subsequent image analysis.

[0056] After treatment, the sample exhibited uniform fluorescence under UV excitation, and the fluorescence signal showed a spatial correspondence with the zeolite distribution. Image processing enabled visualized and quantitative analysis of the zeolite distribution characteristics. Experiments showed a good correlation between fluorescence intensity and zeolite loading in the standard sample (R²>0.98).

[0057] This embodiment illustrates how to perform fluorescence imaging at a specific excitation wavelength after completing the fluorescence labeling of zeolite composite material samples in order to obtain raw optical images for loading and distribution analysis.

[0058] Excitation source: LED ultraviolet exciter, with a main peak wavelength of 490 nm and a bandpass filter width of ±10 nm;

[0059] Sample fixation device: standard glass stage, sample thickness controlled at 2–3 mm;

[0060] Imaging equipment: CMOS fluorescence microscopy imaging system, equipped with a bandpass fluorescence emission filter with a center wavelength of 520 nm;

[0061] Environmental conditions: Operate in a dark room free from natural light interference, maintaining a temperature of 20–25 °C.

[0062] Sample preparation: Take the fluorescently labeled and dried composite material sample and spread it evenly on the black reflective inhibition substrate;

[0063] Excitation settings: Place the sample under the irradiation path of the excitation light source, turn on the excitation light source and adjust the angle and intensity so that the light spot evenly covers the entire sample surface;

[0064] Image acquisition: Turn on the fluorescence microscope imaging system, adjust the focus and set the exposure time (usually 200–500 ms) to capture the fluorescence image of the sample;

[0065] Image output: Save the acquired image as a 16-bit grayscale TIFF format with a resolution of no less than 2048×2048 for subsequent image analysis and quantitative light intensity processing.

[0066] When fluorescently labeled zeolite is excited at 490 nm, its covalently bound fluorescent ligands emit characteristic green light, with a central emission wavelength of approximately 520 nm. Because the zeolite-rich regions contain numerous ligand binding sites, their local fluorescence intensity is enhanced; therefore, the image grayscale is correlated with the spatial distribution of zeolite.

[0067] This imaging method eliminates the need for an electron microscope, relying solely on a standard fluorescence illumination system to acquire large-field-of-view, high-contrast optical images, significantly simplifying equipment requirements. Experiments demonstrate that the mean grayscale value of the imaging region exhibits a non-linear positive correlation with the zeolite mass fraction in the corresponding region, providing a basis for establishing a quantitative model.

[0068] This embodiment provides a method for processing fluorescence images obtained from excitation and generating continuous three-dimensional light intensity spectra, which is applicable to the spatial analysis of zeolite distribution in zeolite composite materials.

[0069] The fluorescence image obtained in the example is in TIFF format and has a size of 2048×2048 pixels.

[0070] The OpenCV library in Python is used to read the grayscale values ​​of an image as a two-dimensional array, with the pixel grayscale range being 0–65535 (16 bits).

[0071] The grid is divided according to the set spatial resolution (e.g., each unit is 100×100 pixels) to generate two-dimensional grid units of the same size.

[0072] The light intensity information within each grid cell is stored as a set of pixel grayscale values ​​to prepare for subsequent processing.

[0073] Calculate the local variance and mean of the grayscale values ​​of all pixels within each grid cell.

[0074] Using two-dimensional grids as X–Y plane data and locally normalized light intensity mean as initial Z-axis value, a sparse three-dimensional point cloud is established.

[0075] A convolutional nested spline interpolation model is introduced, which first performs one-dimensional spline interpolation in each direction, and then performs continuous smoothing through a three-dimensional convolutional kernel;

[0076] This model combines local image gradients with global fitting curves to improve reconstruction accuracy while suppressing abrupt changes.

[0077] The final output is a set of three-dimensional mesh structures, with each voxel containing a normalized light intensity value, forming a continuous and visualized three-dimensional light intensity distribution map.

[0078] Compared with traditional two-dimensional projection or simple interpolation algorithms, this method is more continuous in spatial structure recovery and is suitable for describing the non-uniform distribution characteristics of zeolite along the thickness direction in composite materials, providing high-resolution structural input data for subsequent load estimation and uniformity calculation.

[0079] This embodiment provides a method combining image processing and spatial statistical analysis to evaluate the uniformity of zeolite distribution in composite materials. The method uses three-dimensional fluorescence intensity spectra as data foundation, establishes a zeolite distribution variation coefficient by integrating multiple spatial feature dimensions, and determines whether the zeolite distribution in the material is uniform by comparing this coefficient with a preset threshold.

[0080] First, the three-dimensional light intensity map obtained through excitation light imaging and image processing algorithms is used as input data. This map has been processed by normalization, noise removal, and continuity reconstruction, and possesses complete spatial distribution information.

[0081] To perform spatial analysis, the entire 3D map was divided into several sub-regions of equal volume. Each sub-region corresponds to a local spatial unit, representing the distribution of zeolite at a specific location. This partitioning method allows for a more refined classification of local light intensity within the composite material, which is beneficial for subsequent statistical processing.

[0082] Within each sub-volume region, the light intensity signals of all voxels contained therein are extracted, and these signals are integrated. The integration result can be interpreted as an estimated total zeolite loading within that region, reflecting the local cumulative distribution of zeolite.

[0083] By collecting the integral values ​​of light intensity from all regions, a spatial load distribution map of the entire composite material sample is obtained. This map will be used for subsequent distribution characteristic modeling and analysis.

[0084] After obtaining the light intensity integrals for all sub-regions, a spatial statistical model was further constructed based on these integral values. This model primarily focuses on two aspects of distribution characteristics:

[0085] Skewness: used to reflect whether zeolite has spatial enrichment or biased distribution, such as concentrated loading in some areas and sparse loading in other areas;

[0086] Kurtosis: Used to measure whether the distribution of zeolite is highly concentrated or extreme, indicating that the degree of zeolite accumulation in some areas is far above the average level.

[0087] Both parameters are derived from the statistical characteristics of spatial distribution, which helps to quantitatively capture the "invisible" distribution patterns, and are especially suitable for evaluating the distribution state of zeolites in complex material systems.

[0088] Considering that zeolite may exhibit an asymmetric distribution in a specific direction (such as forming accumulations along the flow direction), this embodiment further introduces an anisotropic analysis mechanism.

[0089] A spherical scanning window with a fixed radius is set in the 3D spectrum, and the window is gradually slid along the x, y, and z axes, recording the mean local fluorescence intensity at each position. Subsequently, the degree of mean change during the window sliding process is calculated to capture distribution fluctuations in different directions.

[0090] This method can reveal the direction dependence of zeolite distribution in composite materials, that is, whether zeolite is more likely to form aggregation or sparse phenomena in a certain direction, thereby judging the structural consistency and stability of the material.

[0091] This embodiment combines the three independent parameters mentioned above—namely, the skewness and kurtosis of the integral distribution of light intensity and the anisotropy distribution index in three dimensions—to construct a composite index that reflects the overall distribution characteristics of zeolite, namely, the zeolite distribution variation coefficient.

[0092] The design concept of this coefficient is to characterize the spatial aggregation trend, anomalous concentration, and directional non-uniformity of zeolites from multiple dimensions. By setting appropriate weights for each parameter, they are linearly combined into a standardized numerical index, which is convenient for use in subsequent discrimination logic.

[0093] A higher ZDVC indicates a more uneven spatial distribution of zeolite, which may indicate problems such as over-enrichment, regional vacancies, or banded arrangement; a lower ZDVC indicates a more balanced and stable distribution and good quality control.

[0094] To enable engineering applications and automatic identification, this embodiment further sets an empirically determined preset threshold. This threshold is derived from comparative experiments of multiple standard samples and expert evaluation results, and is adjusted in conjunction with the quality requirements in the actual preparation process.

[0095] The judgment logic is as follows:

[0096] If the calculated coefficient of variation of zeolite distribution is higher than the preset threshold, it indicates that there is a significant uneven distribution of zeolite in the sample, which may affect its adsorption or catalytic performance, and is judged as "uneven distribution".

[0097] If the coefficient of variation is less than or equal to the threshold, the zeolite in the sample is considered to be evenly distributed, which meets the quality or process requirements and is judged as "uniformly distributed".

[0098] This logic can be automatically completed by the judgment module embedded in image processing software or detection system, improving the efficiency and objectivity of the judgment.

[0099] This embodiment provides a method for constructing a zeolite loading prediction model based on statistical learning. This method establishes a nonlinear fitting relationship between fluorescence intensity signals and zeolite mass content, thereby converting image information of unknown samples into quantifiable zeolite loading data.

[0100] First, several groups of composite material samples with known zeolite mass fractions were selected as standard samples for training the model. The zeolite loading of these standard samples was accurately measured using traditional thermogravimetric analysis or chemical titration, representing typical distributions of different loading levels.

[0101] Subsequently, each of these standard samples was fluorescently labeled. The type of fluorescent ligand, reaction conditions, and image acquisition process used were completely consistent with those of the actual test samples to ensure that the input data for model training was consistent and comparable.

[0102] For each sample, fluorescence images were acquired under the same imaging conditions, and the average gray value in the images was extracted as its corresponding "mean fluorescence intensity". This gray value was normalized and used for subsequent modeling.

[0103] To improve the stability and generalization ability of the model, the collected standard sample data were cleaned. First, outlier samples that might be affected by abnormal excitation, background interference, or positioning shifts during the experiment were removed. Then, the retained data underwent a uniform channel normalization operation to ensure that the fluorescence intensity values ​​of different batches and samples were comparable within the same numerical range.

[0104] Normalization helps reduce systematic errors, enhances the model's tolerance to different sample inputs, and ensures that subsequent fitted models more accurately reflect real physical relationships.

[0105] To accurately describe the nonlinear relationship between fluorescence intensity and zeolite loading, this embodiment uses a locally weighted regression method for modeling.

[0106] The core idea of ​​this method is to select a set of samples that are close in fluorescence intensity space near each input value, forming a local data subset, and then fit a simple regression model, such as a first-order or second-order polynomial, onto this subset. In this way, the entire input interval is covered by multiple local regression results, which are then weighted and concatenated to form a continuous global fitting curve.

[0107] Compared with traditional global linear or high-order polynomial fitting methods, local weighted regression has higher flexibility and accuracy, and is especially suitable for situations where the data distribution is non-uniform or the response relationship varies greatly. It can effectively avoid underfitting or overfitting.

[0108] In practice, for each standard sample point, a set of neighboring points in the fluorescence intensity space is dynamically selected, and a curve is fitted to this local region separately. The model assigns a weight to each sample point, which decreases as the sample point is closer to the target point in the input space, thereby enhancing local correlation and weakening long-distance interference.

[0109] During the modeling process, to ensure optimal predictive performance, the parameters of the locally fitted model are optimized and adjusted, such as the neighborhood size, regression order, and weighting function form. This optimization is achieved through a cross-validation strategy, which involves randomly dividing all standard samples into multiple subsets, which alternately serve as the training and validation groups to evaluate the model's predictive ability on unseen data.

[0110] Finally, using the minimum average error as the objective function, the parameter combination with the best fitting accuracy was selected as the final model structure, thus completing the construction of the nonlinear prediction model for zeolite loading.

[0111] The three-dimensional light intensity distribution spectrum of the composite material sample under test was obtained through fluorescence imaging and image processing. The spectrum was then subjected to channel normalization to ensure that the grayscale range was consistent with the training model.

[0112] The spectrum was segmented into multiple voxel regions (i.e., three-dimensional pixel units), and the average gray value of each voxel was extracted as its representative fluorescence intensity input. For each voxel, its gray value was substituted into the previously constructed fitting curve to obtain the corresponding predicted zeolite mass fraction value.

[0113] After prediction, the zeolite mass fractions of all voxels are volume-weighted and integrated to output the average loading of the entire sample. This weighting method not only considers the predicted value of each voxel but also reflects its spatial proportion in the overall sample structure, thereby improving the accuracy of the overall estimation.

[0114] Example 2: To verify the feasibility and superiority of the detection method provided by this invention in practical applications, five groups of composite material standard samples with different zeolite loadings (5%, 10%, 15%, 20%, and 25%) were selected for testing. The zeolite mass fraction was detected using the fluorescence image analysis method of this invention and the traditional thermogravimetric analysis (TGA), and the distribution was evaluated and compared using scanning electron microscopy images combined with energy dispersive spectroscopy (EDS).

[0115] Sample preparation: A fixed mass of ZSM-5 zeolite was mixed with γ-alumina powder in a certain proportion, pressed into tablets, and sintered to prepare composite material sheets with known loading.

[0116] Test procedure for the method of this invention:

[0117] Fluorescent labeling: 4-Carboxyfluorescein was used as the fluorescent ligand, and the solution treatment time was 2 hours;

[0118] Fluorescence imaging: excitation at 490 nm, acquisition of fluorescence images and image normalization;

[0119] 3D map construction: based on image rasterization and spline interpolation;

[0120] Load estimation: Prediction is based on a locally weighted regression model;

[0121] Uniformity assessment: The coefficient of variation (ZDVC) of zeolite distribution is calculated based on the integral distribution of light intensity and the anisotropy index.

[0122] Traditional testing process:

[0123] The TGA method was used to measure the mass loss of the sample at 600 °C, and the zeolite content was calculated.

[0124] SEM-EDS was used to obtain elemental distribution maps of zeolite regions, and the uniformity of distribution was subjectively assessed.

[0125]

[0126] As shown in Table 1, the product composition is the mass percentage of zeolite and alumina in each sample; the predicted values ​​of the method of this invention are obtained by estimating the image grayscale and fitting model; the measured values ​​of TGA are obtained by thermogravimetric analysis; ZDVC is the quantitative index used in this invention to characterize the spatial distribution uniformity of zeolite, and the larger the value, the more uneven the distribution; the SEM observation conclusions are derived from the subjective distribution judgment of scanning electron microscopy-energy dispersive spectroscopy imaging and are used for auxiliary comparison.

[0127] Performance Analysis: In the five groups of samples, the maximum deviation between the method of this invention and the TGA measured results did not exceed ±0.4%, indicating that the established nonlinear mapping model between fluorescence intensity and zeolite loading has a good fitting effect. The method of this invention takes about 15 minutes from image acquisition to result output, which is significantly better than the 2-3 hours required by TGA. The coefficient of variation (ZDVC) index enables a quantitative assessment of distribution uniformity, and can clearly identify phenomena such as boundary accumulation and center agglomeration. Traditional SEM methods can only be used for qualitative observation and are difficult to quantify. In addition, the entire processing flow supports software execution, with less manual intervention, avoiding subjective errors, and is suitable for batch detection scenarios.

[0128] The experimental results of this embodiment show that the fluorescence image analysis and regression modeling detection method proposed in this invention has high precision, fast response and automation capabilities in zeolite loading determination. At the same time, by constructing a spatial variation coefficient index, it realizes the quantitative judgment of distribution uniformity, which is significantly better than existing traditional methods such as thermogravimetric analysis and microscopic observation.

[0129] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A rapid detection method for zeolite loading and uniformity in zeolite composite materials, characterized in that: include: The zeolite composite material sample to be tested is brought into contact with a solution containing fluorescent ligands to form a fluorescent label; Fluorescence imaging of the sample is performed under selected wavelength excitation to obtain an optical image of the marked area; Using image processing algorithms, the acquired optical images are rasterized and a three-dimensional light intensity distribution map is constructed; Based on the three-dimensional light intensity distribution map, the coverage of zeolite and the mean deviation of light intensity in each region are quantified, and the zeolite distribution variation coefficient is output to evaluate the uniformity of zeolite distribution in the composite material. The calculation of the zeolite distribution variation coefficient includes: based on the three-dimensional light intensity distribution map, dividing the map into multiple sub-volume regions according to spatial hierarchy, and extracting the fluorescence intensity integral value of zeolite in each sub-region; constructing a probability density distribution model of zeolite light intensity in the overall space, and calculating its distribution skewness and kurtosis; introducing a spherical scanning window structure, moving the window in each direction to statistically analyze the local mean change rate, forming an anisotropic distribution index; combining the fluorescence intensity integral value, distribution skewness and kurtosis, and anisotropic distribution index, outputting the zeolite distribution variation coefficient through a multi-factor weighted model as the zeolite uniformity detection result; The obtained coefficient of variation of zeolite distribution is compared with a preset threshold. If the coefficient of variation of zeolite distribution is greater than the preset threshold, it indicates that the distribution of zeolite in the composite material is uneven; if the coefficient of variation of zeolite distribution is less than or equal to the preset threshold, it indicates that the distribution of zeolite in the composite material is uniform. A fitting curve was established between fluorescence signal intensity and standard zeolite loading. Based on the fitting curve, the image grayscale value was correlated with the actual zeolite mass content, and the zeolite loading was estimated.

2. The rapid detection method for zeolite loading and uniformity in a zeolite composite material according to claim 1, characterized in that: The image processing algorithm includes: The optical fluorescence image obtained by excitation is processed into a two-dimensional pixel matrix, and the image area is divided into grids according to a preset spatial resolution to construct a light intensity data raster unit. Local variance normalization is performed on the light intensity value of each pixel in each grid cell; A convolutional nested spline interpolation model is used to perform longitudinal interpolation fitting on the light intensity data of each grid cell to generate a continuous three-dimensional light intensity distribution map.

3. The rapid detection method for zeolite loading and uniformity in a zeolite composite material according to claim 1, characterized in that: The establishment of the fitting curve between the fluorescence signal intensity and the standard zeolite loading includes: Multiple sets of composite material standard samples with known zeolite mass fractions were selected, and their corresponding average fluorescence intensity values ​​were obtained through the same fluorescent labeling and image acquisition process; Outlier removal and light intensity signal normalization were performed on the collected data; Based on a local weighted regression model, the nonlinear mapping relationship between fluorescence intensity and actual zeolite loading is fitted. Model parameters were optimized using cross-validation and the minimum mean square error criterion.

4. The rapid detection method for zeolite loading and uniformity in a zeolite composite material according to claim 3, characterized in that: The establishment of the nonlinear mapping relationship is based on a local weighted regression model, specifically including: For each standard sample point, a local neighborhood sample subset is dynamically selected based on the position of its fluorescence intensity value in the input space; In each local region, a first-order or second-order polynomial regression model is independently fitted to obtain the estimated values ​​of fluorescence intensity and zeolite loading in that region. Multiple local regression results are combined to form a global fitting curve.

5. The rapid detection method for zeolite loading and uniformity in a zeolite composite material according to claim 4, characterized in that: The estimated zeolite loading includes: The three-dimensional light intensity distribution map obtained through image processing is channel normalized, and the average gray value of each grid voxel region is extracted; The average gray value is used as an input variable and substituted into the fitting relationship curve constructed based on the local weighted regression model to obtain the corresponding predicted value of zeolite mass fraction. The predicted values ​​for all voxel regions are volume-weighted averaged to calculate the total zeolite loading of the entire sample.

Images