A hydrophobic environmentally friendly coating and its preparation method
By optimizing the use of coupling agents and crosslinking agents through dynamic mechanical analysis, orthogonal experiments, and response surface methodology, the problem of performance imbalance in waterborne coatings in industrial applications was solved, achieving simultaneous improvement in coating adhesion, hydrophobicity, and hardness, as well as enhanced batch stability.
Patent Information
- Application Number
- CN202611039469.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-14
- Publication Date
- 2026-11-17
AI Technical Summary
Existing water-based environmentally friendly coatings face problems such as slow film drying speed, insufficient adhesion, poor water resistance, and poor chemical resistance in industrial applications. Furthermore, the selection of coupling agents lacks quantitative testing, the particle size combination of fillers has not been systematically compared, and the screening of crosslinking agents has failed to balance hydrophobicity and hardness, resulting in performance imbalance.
The optimal coupling agent type and concentration were screened using dynamic mechanical analysis, particle size combinations were designed using orthogonal experimental design, the optimal particle size was screened using range analysis, and the crosslinking agent dosage equation was constructed using response surface methodology. The use of coupling agent and crosslinking agent was optimized through Gaussian process regression and Bayesian update to ensure the synchronous improvement of coating performance.
This achieved simultaneous improvement in coating adhesion, hydrophobicity, and hardness, shortening the development cycle, enhancing batch stability, and reducing material waste.
Smart Images

Figure CN123120130A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of coating technology, specifically to a hydrophobic environmentally friendly coating and its preparation method. Background Technology
[0002] With increasingly stringent environmental regulations, traditional solvent-based coatings are facing gradual elimination due to volatile organic compound (VOC) emissions, making water-based coatings and other environmentally friendly coatings the mainstream direction for industry development. Simultaneously, endowing coatings with hydrophobic and self-cleaning properties has become an important development trend in coating technology. However, existing water-based environmentally friendly coatings still face many bottlenecks in industrial applications: the high latent heat of vaporization of water leads to slow film drying; the large number of hydrophilic groups introduced into the resin structure makes it difficult to meet the requirements of industrial applications in terms of film hardness, water resistance, and chemical resistance; at the same time, existing water-based coatings generally suffer from insufficient adhesion and poor water resistance, especially in harsh environments such as marine engineering and petrochemicals, resulting in a service life far shorter than that of traditional solvent-based coatings. How to simultaneously improve the hydrophobicity, adhesion, and hardness of coatings while ensuring environmental friendliness has become a pressing technical challenge in this field.
[0003] Current technologies rely heavily on supplier recommendations or empirical formulations for coupling agent selection, lacking quantitative methods to test their actual bonding strength with fillers. Improper selection can easily lead to weak interfacial bonding and insufficient coating adhesion. Filler particle size combinations are often adjusted using single factors or fixed ratios, neglecting the synergistic effect of micron and nanoparticles on adhesion. In actual production, different particle size ratios alter packing density and stress distribution; without systematic comparison, micro-cracks can easily occur. Crosslinking agent screening focuses only on hardness or hydrophobic angle, ignoring the competitive relationship between the two, resulting in performance imbalances. Crosslinking agent dosage relies on manually preparing gradient samples and trial-and-error; if the testing range deviates from the optimal range, precise dosage cannot be determined, leading to large batch-to-batch performance fluctuations and significant material waste. Overall optimization methods lack experimental design and mathematical model support; the interrelationships between parameters remain unresolved, development time often exceeds several months, and it is difficult to simultaneously meet multiple standards such as hydrophobicity, adhesion, and hardness. Summary of the Invention
[0004] The purpose of this invention is to overcome the shortcomings of the existing technology and to propose a hydrophobic environmentally friendly coating and its preparation method.
[0005] The technical solution provided by this invention to solve the above problems is: a method for preparing a hydrophobic and environmentally friendly coating, characterized in that the preparation method includes the following steps:
[0006] S1. The bonding strength between each silane coupling agent and the filler is obtained by dynamic mechanical analysis, and the bonding strength values are compared to screen out the best coupling agent type. Then, a stratified gradient concentration experiment is carried out to obtain bonding force data, and the best coupling agent concentration is screened out by comparison.
[0007] S2. Based on the optimal coupling agent type and optimal coupling agent concentration as fixed parameters, an orthogonal experimental design was used to arrange experiments on particle size combinations of micron-sized and nano-sized fillers to obtain adhesion data for each combination. Range analysis was performed to compare and select the optimal particle size combination.
[0008] S3. Based on the optimal coupling agent type and optimal coupling agent concentration as fixed parameters, multiple crosslinking agents are selected to prepare coatings in the same proportion. After detecting the hydrophobic angle value and hardness value, the values are normalized and then the weighted sum is calculated. The weighted sum values are compared to screen out the optimal crosslinking agent type.
[0009] S4. Based on the optimal type of crosslinking agent as the variable benchmark, and with the optimal coupling agent type and optimal coupling agent concentration as fixed parameters, a quadratic regression equation between adhesion and crosslinking agent dosage is constructed using the response surface methodology. The dosage value corresponding to the extreme point of the quadratic regression equation is calculated by taking the derivative of the quadratic regression equation as the optimal crosslinking agent dosage.
[0010] S5. Preparation of hydrophobic environmentally friendly coatings based on optimal coupling agent concentration, optimal particle size combination, optimal crosslinking agent type, and optimal crosslinking agent dosage.
[0011] 2. The method for preparing a hydrophobic environmentally friendly coating according to claim 1, characterized in that, in step S1, the bonding strength between each silane coupling agent and the filler is obtained by dynamic mechanical analysis, and the optimal coupling agent type is selected by comparing the bonding strength values. Specifically, this includes:
[0012] Data on the molecular weight, polarity, mixing ratio, temperature, and humidity characteristics of candidate silane coupling agents, as well as data on filler particle size and surface energy characteristics, are obtained. These data are then input into a Gaussian process regression model for training to output a dynamic simulation model.
[0013] Three dynamic mechanical analysis measurements were performed on each candidate silane coupling agent and filler under the conditions of 25 degrees Celsius and 50% humidity. The measured binding strength data of each measurement were collected and the average value of the three measurements was calculated as the average binding strength. The average binding strength, along with the coupling agent identification and experimental conditions, was packaged into a measured feedback dataset.
[0014] The dynamic simulation model and the measured feedback dataset are called. The hyperparameters of the dynamic simulation model are iteratively corrected using the Bayesian update algorithm. The average binding strength in the measured feedback dataset is used as the observation value to update the posterior distribution of the prediction function, and the corrected dynamic simulation model is output.
[0015] The characteristic data of all candidate silane coupling agents are input into the modified dynamic simulation model for prediction. The predicted binding strength value of each coupling agent is obtained. The coupling agent type corresponding to the first position in the ranking is selected as the best coupling agent type.
[0016] Five gradient concentration levels (0.5%, 1.0%, 1.5%, 2.0%, and 2.5% of filler mass) were designed based on the optimal coupling agent type. For each group of three samples, the bonding strength was measured and the average value was calculated. A modified dynamic simulation model was used to predict the bonding strength. The measured average value and the predicted value were weighted and summed with weights of 0.6 and 0.4, respectively. The concentration corresponding to the maximum weighted sum was selected as the optimal coupling agent concentration. This scheme utilizes a Gaussian process regression model to construct a dynamic simulation model, providing measured bonding strength data. A Bayesian update algorithm is used to feed the measured data back to the model for iterative correction, allowing the model's predictive ability for bonding strength to continuously improve with the accumulation of experimental data, avoiding the limitations of a single static evaluation. Based on the modified dynamic simulation model, all candidate coupling agents are predicted and ranked, and the optimal coupling agent type is selected. A weighted sum of the measured data from the stratified gradient concentration experiments and the model predictions is then used to scientifically select the optimal coupling agent concentration from the five concentration levels. The entire process achieves data-driven dynamic optimization, enhancing the adaptability and reliability of coupling agent screening, and providing a more accurate basis for determining filler modification schemes in subsequent coating preparation.
[0017] Preferably, the step S2, which involves range analysis and comparison to select the optimal particle size combination, specifically includes the following steps:
[0018] By collecting adhesion data, performing mean normalization, calculating the difference between each data point and the mean and dividing it by the standard deviation, standardized adhesion data is obtained. Pearson correlation coefficients between each parameter and the standardized adhesion data are calculated, and the top four are selected as dominant features based on the magnitude of the correlation coefficients.
[0019] Based on the dominant features and standardized adhesion data, principal component analysis is used to calculate the covariance matrix. The top two principal components with an eigenvalue ratio of more than 85% are selected to construct a dimension-reduced feature matrix. The dimension-reduced feature matrix and standardized adhesion data are input into a self-organizing map network. The winning node is determined by Euclidean distance calculation and the weights are adjusted. When the weight update magnitude exceeds twice the standard deviation, an update is performed. The group label vector and the dynamically adjusted standardized adhesion data are output.
[0020] For the dynamically adjusted standardized adhesion data, the sparse data part is identified based on the sparsity threshold. The corresponding group label vector is input into the generative adversarial network. The generator uses a multilayer perceptron to generate the complete data. The discriminator calculates the error based on the cross-entropy loss function. When the loss value is lower than 0.01, the completed adhesion data is output. The completed adhesion data and the dynamically adjusted standardized adhesion data are merged into a comprehensive adhesion dataset.
[0021] Based on the comprehensive adhesion dataset, the average value and range of the adhesion data corresponding to each particle size combination are calculated to obtain the set of range values; Based on the range set, each particle size combination is sorted in ascending order according to its range, and the combination with the smallest range is selected as the optimal particle size combination. In the above scheme, mean normalization and Pearson correlation coefficient are introduced sequentially to select dominant features, principal component analysis is used for dimensionality reduction, self-organizing map network is used for dynamic clustering and weight adjustment, and generative adversarial network is used to complete sparse data before finally calculating the range and sorting. This series of data processing methods significantly improves the quality of adhesion data—standardization eliminates the influence of dimensions, feature selection and dimensionality reduction remove redundant information, clustering and dynamic adjustment enhance the data's discriminative power, and generative adversarial network fills in missing values in sparse regions. This effectively avoids misjudgments caused by data noise, insufficient samples, or dimensionality interference in the original orthogonal experiment, greatly enhancing the scientific rigor, robustness, and accuracy of the particle size combination selection process. The final optimal particle size combination more accurately reflects the comprehensive influence of filler particle size on adhesion, providing more reliable basic parameters for subsequent crosslinking agent selection, dosage optimization, and coating preparation, ultimately improving the optimization efficiency of the overall preparation method and the stability of coating performance.
[0022] Preferably, S3 specifically includes:
[0023] Using the optimal coupling agent type and optimal coupling agent concentration as fixed parameters, several crosslinking agents were selected and mixed with the coupling agent in the same proportion to prepare coatings. The hydrophobic angle values of each coating were recorded using a contact angle measuring instrument, and the hardness values of each coating were recorded using a hardness tester to generate a hydrophobic angle hardness dataset.
[0024] Based on the hydrophobic angle and hardness dataset, the hydrophobic angle value and hardness value are mapped to intervals using the maximum and minimum value normalization method. For each crosslinking agent, the normalized hydrophobic angle value and hardness value are weighted and summed according to preset weights to obtain the comprehensive performance score set.
[0025] For the comprehensive performance score set, compare the scores of all crosslinking agents, select the crosslinking agent with the highest score, record the crosslinking agent number, and obtain the optimal crosslinking agent type.
[0026] Preferably, S4 specifically includes:
[0027] The optimal type of crosslinking agent was used as the baseline variable, the amount of crosslinking agent was set as the independent variable, three incremental amount levels were set, the experimental points were arranged according to the central composite design principle, and an amount level table was generated.
[0028] Based on the dosage level table, the coating slurry was prepared and applied to the standard substrate. After curing, the adhesion tester was used to measure the adhesion point by point according to the standard method. The adhesion readings corresponding to each dosage level were recorded and compiled into an adhesion data matrix.
[0029] Based on the adhesion data matrix, a quadratic polynomial equation is fitted using the least squares method to obtain the regression coefficient of adhesion with respect to the amount of crosslinking agent. The first derivative of the regression equation is calculated, and the derivative is set to zero to solve for the amount of crosslinking agent at the extreme point, thus obtaining the optimal amount of crosslinking agent.
[0030] Preferably, S5 specifically includes:
[0031] The resin mass baseline is 100 parts. The filler mass corresponding to the optimal particle size combination is set as a fixed proportion of the resin mass. The filler mass parts are generated by multiplying the resin mass baseline by this proportion.
[0032] Based on the filler mass fraction, the optimal coupling agent concentration value is divided by the filler mass fraction and then multiplied by 100 to obtain the relative filler mass percentage. This percentage is then multiplied by the filler mass fraction. Simultaneously, the optimal crosslinking agent dosage value is divided by the resin mass reference and then multiplied by 100 to obtain the relative resin mass percentage. This percentage is then multiplied by the resin mass reference to obtain the coupling agent mass fraction and crosslinking agent mass fraction.
[0033] The mass fractions of filler, coupling agent, and crosslinking agent are called. Based on 100 parts of resin, the three are standardized according to their corresponding standard amounts. The proportions are combined to obtain the respective parts and generate the component values of the formulation.
[0034] According to the formula component values, the filler, coupling agent, crosslinking agent and other raw materials are added to the mixing tank in sequence and dispersed at high speed, and then transferred to the grinding mill for grinding to obtain the hydrophobic environmentally friendly coating.
[0035] The present invention also discloses a hydrophobic environmentally friendly coating, which is prepared by any of the preparation methods described above.
[0036] Compared with existing technologies, the advantages of this invention are as follows: This invention uses dynamic mechanical analysis to quantify the bonding strength between silane coupling agents and fillers, sorts coupling agent types by numerical values, and supplements this with stratified gradient concentration experiments to obtain bonding force data, ensuring that the selection of coupling agents has a reproducible mechanical basis. Orthogonal experiments are arranged to combine micron and nano-sized filler particles, and range analysis is used to determine the main effects and optimal levels, covering the interactive effects of multiple factors with a small number of experiments. Hydrophobic angle and hardness are measured and then normalized and weighted to eliminate dimensional differences, comprehensively balancing the two properties to screen crosslinking agent types. Response surface methodology is used to construct a quadratic regression equation of adhesion on crosslinking agent dosage, and the derivative is used to directly calculate the extreme point relative to the application amount, avoiding trial-and-error errors. This progressive logic decomposes the multivariate coupling problem into controllable sub-steps, with the results of each stage serving as a fixed benchmark for subsequent steps. This allows the synergistic effect between fillers, coupling agents, and crosslinking agents to be systematically captured, ultimately resulting in simultaneous improvement in coating adhesion, hydrophobicity, and hardness, shortening the development cycle, and enhancing batch stability. Attached Figure Description
[0037] The accompanying drawings, which are provided to further illustrate the invention and constitute a part of this invention, are illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention.
[0038] Figure 1 This is the method flowchart of the present invention. Figure 1 ;
[0039] Figure 2 This is the method flowchart of the present invention. Figure 2 ;
[0040] Figure 3 This is the method flowchart of the present invention. Figure 3 . Detailed Implementation
[0041] The following will describe in detail the implementation of the present invention with reference to the accompanying drawings and embodiments, so that the process of how the present invention uses technical means to solve technical problems and achieve technical effects can be fully understood and implemented accordingly.
[0042] In this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features. Thus, the use of "first" and "second" to define a feature may explicitly or implicitly include one or more of that feature, and in the description of this invention, "a number" means two or more, unless otherwise explicitly specified.
[0043] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.
[0044] It should also be understood that the terminology used in this specification of embodiments of the invention is for the purpose of describing particular embodiments only and is not intended to limit the embodiments of the invention. As used in this specification of embodiments of the invention and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise. Example 1
[0045] like Figure 1 As shown in the figure, this embodiment discloses a method for preparing a hydrophobic environmentally friendly coating, the preparation method including the following steps:
[0046] S1. The bonding strength between each silane coupling agent and the filler is obtained by dynamic mechanical analysis, and the bonding strength values are compared to screen out the best coupling agent type. Then, a stratified gradient concentration experiment is carried out to obtain bonding force data, and the best coupling agent concentration is screened out by comparison.
[0047] S2. Based on the optimal coupling agent type and optimal coupling agent concentration as fixed parameters, an orthogonal experimental design was used to arrange experiments on particle size combinations of micron-sized and nano-sized fillers to obtain adhesion data for each combination. Range analysis was performed to compare and select the optimal particle size combination.
[0048] S3. Based on the optimal coupling agent type and optimal coupling agent concentration as fixed parameters, multiple crosslinking agents are selected to prepare coatings in the same proportion. After detecting the hydrophobic angle value and hardness value, the values are normalized and then the weighted sum is calculated. The weighted sum values are compared to screen out the optimal crosslinking agent type.
[0049] S4. Based on the optimal type of crosslinking agent as the variable benchmark, and with the optimal coupling agent type and optimal coupling agent concentration as fixed parameters, a quadratic regression equation between adhesion and crosslinking agent dosage is constructed using the response surface methodology. The dosage value corresponding to the extreme point of the quadratic regression equation is calculated by taking the derivative of the quadratic regression equation as the optimal crosslinking agent dosage.
[0050] S5. Preparation of hydrophobic environmentally friendly coatings based on optimal coupling agent concentration, optimal particle size combination, optimal crosslinking agent type, and optimal crosslinking agent dosage.
[0051] Specifically, the optimal coupling agent type refers to the type of chemical substance that can effectively enhance the interfacial bonding between the filler and the resin, such as silane coupling agents with different functional groups. Its selection is based on the substance's reactivity and compatibility. The optimal coupling agent concentration refers to the effective concentration range in practical operation; controlling it within this range achieves the best interfacial bond strength. The optimal particle size combination emphasizes the classification and distribution of filler particle sizes, which can bring excellent physical properties, such as better adhesion. The optimal crosslinking agent type focuses on the chemical structure and reaction mechanism of the crosslinking agent, ensuring the chemical stability and mechanical durability of the coating. The comparison between the optimal crosslinking agent dosage and resin quality ensures the improvement and optimization of material performance. The coating formulation is the final product, with optimized raw material quality and proportions to better meet actual performance requirements.
[0052] like Figure 2 As shown, in S1, the bonding strength between each silane coupling agent and the filler is obtained through dynamic mechanical analysis, and the bonding strength values are compared to screen out the optimal coupling agent type. Specifically, this includes:
[0053] The binding strength between various silane coupling agents and fillers was obtained using dynamic mechanical analysis. A database of candidate silane coupling agent characteristics was constructed, collecting molecular weight, polarity, mixing ratio, temperature (degrees Celsius), and humidity (relative humidity percentage) parameters for each coupling agent. Simultaneously, filler particle size and surface energy parameters were also collected. These eight parameters were used as the input feature matrix, with binding strength as the output variable. A Gaussian process regression model was applied, and the covariance matrix was calculated using the radial basis function kernel. Hyperparameters were estimated by maximizing the logarithmic marginal likelihood. After training, a dynamic simulation model was output, with the specific formula predicting the binding strength value. Mean function ,in Let covariance be the vector between test points and training points. To train the covariance matrix of the data, For noise variance, It is the identity matrix. This is a vector of intensity observations combined with the training set.
[0054] Under constant conditions of 25°C and 50% humidity, three standard tensile specimens were prepared for each candidate silane coupling agent and filler combination. Each specimen was subjected to sinusoidal oscillatory loading at a frequency of 1 Hz and a strain amplitude of 0.1% using a dynamic mechanical analyzer. The storage modulus and loss modulus of each specimen were directly collected, and the bond strength was determined using the formula. (in For energy storage modulus, The single measured bond strength value is calculated for the loss modulus. The arithmetic mean of the three specimens is calculated as the average bond strength of the combination. The average bond strength data is combined with the coupling agent number (numerical identifier 1 to N) and experimental condition parameters (temperature 25 and humidity 50) into a triplet vector to form the measured feedback dataset.
[0055] The dynamic simulation model and the measured feedback dataset are called. The hyperparameters of the dynamic simulation model are iteratively corrected using the Bayesian update algorithm. The average binding strength in the measured feedback dataset is used as the observation value to update the posterior distribution of the prediction function, and the corrected dynamic simulation model is output.
[0056] By calling the predicted mean and covariance functions output by the aforementioned dynamic simulation model, and combining them with the average binding strength observations in the measured feedback dataset, a Bayesian update algorithm is used to update the posterior mean of the prior Gaussian process. The posterior covariance is updated to ,in This is the measured average bonding strength vector. Given the value of the prior mean function on the training input, iteratively correct the two hyperparameters, kernel length scale and signal variance, until the posterior likelihood function value converges, and output the corrected dynamic simulation model.
[0057] The test input matrix is composed of seven characteristic data points (each a continuous value) corresponding to all candidate silane coupling agents: molecular weight, polarity, mixing ratio, temperature, humidity, filler particle size, and surface energy. This matrix is then input into the corrected dynamic simulation model, which outputs a predicted binding strength value (in megapascals) for each coupling agent. These values are sorted from largest to smallest, and the coupling agent ranked first is selected. Its coupling agent type identifier is recorded, and this coupling agent is defined as the optimal coupling agent type.
[0058] Based on the optimal coupling agent type mentioned above, five gradient concentration levels were designed: 0.5%, 1.0%, 1.5%, 2.0%, and 2.5% (percentage of filler mass). Three parallel specimens were prepared for each concentration level. Each specimen was subjected to a tensile tester at a tensile rate of 5 mm / min until interface failure. The maximum load was recorded and divided by the bonding area to obtain the bonding force (in Newtons per square millimeter). The arithmetic mean of the three specimens at each concentration was calculated as the average bonding force. Simultaneously, a modified dynamic simulation model was invoked, inputting the concentration and the characteristic data of the same filler, and outputting the predicted bonding force value corresponding to that concentration. A weighted sum was calculated for each concentration. ,in Let i be the average binding force at the i-th concentration. To predict binding force, the maximum value is selected from five weighted sum values, and its corresponding concentration value is recorded as the optimal coupling agent concentration. This concentration is divided into three intervals according to the weighted sum distribution: if the concentration corresponding to the maximum value is lower than 1.0%, it is a low concentration interval; 1.0% to 2.0% is a medium concentration interval; and higher than 2.0% is a high concentration interval.
[0059] Specifically, the step of S1, which involves conducting a stratified gradient concentration experiment to obtain binding force data and comparing and selecting the optimal coupling agent concentration, includes: conducting a stratified gradient concentration experiment on the optimal coupling agent type, recording the binding force data at each concentration, and selecting the coupling agent concentration corresponding to the maximum value of the binding force data as the optimal coupling agent concentration by comparing the binding force data at each concentration.
[0060] The above scheme integrates dynamic simulation and experimental feedback using Bayesian methods, which utilizes the model's global prediction capabilities and corrects them with experimental data. This ensures that the selection of the optimal coupling agent type and concentration has both theoretical guidance and experimental basis. At the same time, the concentration selection is processed by weighted comprehensive balance to avoid the bias of a single data source, ultimately improving the efficiency and reliability of interface-based optimization.
[0061] The dynamic simulation model includes a prediction function, a kernel function, and hyperparameters; the measured feedback dataset includes average binding strength, coupling agent identification, temperature, and humidity; and the corrected dynamic simulation model includes updated hyperparameters and posterior prediction distribution.
[0062] like Figure 2 As shown, in S2, based on the optimal coupling agent type and optimal coupling agent concentration as fixed parameters, an orthogonal experimental design was used to arrange experiments on particle size combinations of micron-sized and nano-sized fillers to obtain adhesion data for each combination. Specifically, this included:
[0063] Based on the optimal coupling agent type and optimal coupling agent concentration as fixed parameters, an orthogonal experimental design was used to arrange particle size combination experiments of micron-sized and nano-sized fillers. Micron-sized fillers (e.g., particle size specifications including 5 microns, 10 microns, and 20 microns) and nano-sized fillers (e.g., particle size specifications including 20 nanometers, 50 nanometers, and 80 nanometers) were selected, with three levels for each level, forming a total of nine particle size combinations as experimental factors. Nine experimental schemes were arranged according to the L9(3²) orthogonal array. One coating sample was prepared for each scheme. After curing, the maximum force (in Newtons per square millimeter) of the coating peeling off from the substrate was measured using an adhesion tester with the pull-off method at a loading rate of 10 mm per minute. The data were recorded as adhesion data and formed a nine-element vector.
[0064] For example, the process of performing range analysis in S2 to screen for the optimal particle size combination specifically includes the following steps:
[0065] The adhesion data above were normalized to the mean. First, the arithmetic mean of the nine adhesion data sets was calculated. and standard deviation For each adhesion data point Calculate standardized values The standardized adhesion data vector was obtained; the Pearson correlation coefficient between each parameter (micron particle size, nano particle size and their interaction terms) and the standardized adhesion was calculated. ,in Let p be the value of the p-th parameter in the j-th group. Given the mean of this parameter, the four parameters with the largest absolute values of correlation coefficients are selected from the five candidate parameters as the dominant features.
[0066] Based on four dominant characteristics and standardized adhesion data, a design matrix is constructed. (Nine rows and four columns), for Calculate the covariance matrix ,right Eigenvalues are obtained by performing eigenvalue decomposition. According to the formula, the first k principal components are selected to maximize the cumulative contribution rate. Take the corresponding eigenvectors to form the projection matrix (4 rows, 2 columns) Calculate the feature matrix after dimensionality reduction. (Nine rows and two columns); will With standardized adhesion data vector Input a self-organizing map network with a grid size of 3x3 and nine neurons. Calculate the Euclidean distance between each input sample and the weights of each neuron. Select the neuron with the smallest distance as the winning node. Update the weights of the winning node and its neighboring regions based on a learning rate of 0.1. Calculate the standard deviation of the weight update magnitude after each iteration. Stop the iteration when the update magnitude exceeds twice the current standard deviation. Output the group label vector (nine-dimensional) and the dynamically adjusted standardized adhesion data (vector form).
[0067] For the dynamically adjusted standardized adhesion data, a sparsity threshold of 0.15 is set. This means that when the mean Euclidean distance between a data point and its three nearest neighbors is greater than 1.5 times the overall mean distance, it is considered sparse data. The identified sparse data points are grouped into labels. The sparse group label vectors are input into a generative adversarial network (GAN). The generator uses a three-layer multilayer perceptron (2 input neurons, 64 hidden neurons, and 1 output neuron) to generate complete data. The discriminator uses a two-layer multilayer perceptron (1 input neuron, 32 hidden neurons, and 1 output neuron), with a cross-entropy loss function. Calculation error (m is the batch size, For true labels, 0 or 1, (To predict probabilities), iterative training stops when the loss value is below 0.01, outputting complete adhesion data, and merging the complete data with the dynamically adjusted standardized adhesion data into a comprehensive adhesion dataset (with nine complete sets of data in the dimension).
[0068] Based on the comprehensive adhesion dataset, the mean adhesion value was calculated for each particle size combination (nine groups in total). (Set of nine means) and range (Nine sets of range values) Obtain the set of range values; sort the nine sets of range values from largest to smallest. The larger the range value, the greater the adhesion fluctuation and the poorer the process stability of the combination. The smaller the range value, the better the adhesion consistency. Select the combination with the smallest range value and record it as the optimal particle size combination. This combination includes two values: micron particle size and nano particle size. The particle size distribution structure of the optimal particle size combination is divided into continuous gradation (micron to nano ratio between 0.5 and 2.0) and discontinuous gradation (ratio outside this range) according to the ratio of the two particle sizes. The packing density characteristic is judged by calculating the packing density (total mass of filler divided by apparent volume). A density greater than 1.5 grams per cubic centimeter is considered dense, and a density lower than this is considered loose.
[0069] The above scheme comprehensively utilizes orthogonal design, correlation analysis, principal component dimensionality reduction, self-organizing map clustering, generative adversarial network completion, and range analysis to construct a complete process from experimental arrangement to data cleaning and feature selection. It can efficiently identify the particle size combination that minimizes adhesion fluctuations with a limited number of experiments, and significantly improve the robustness of coating preparation.
[0070] The standardized adhesion data includes the z-score value of each data point; the dominant feature includes four feature parameter indices; the dimensionality-reduced feature matrix includes the first principal component score vector and the second principal component score vector; the group label vector includes the cluster label of each sample; the dynamically adjusted standardized adhesion data includes the updated adhesion normalized value; the completed adhesion data includes the generated adhesion value; the comprehensive adhesion dataset includes the dynamically adjusted standardized adhesion data and the completed adhesion data; and the range set includes the range values of each particle size combination.
[0071] For example, S3 specifically includes:
[0072] Using the optimal coupling agent type and concentration as fixed parameters, five crosslinking agents (numbered C1 to C5) were selected. Each crosslinking agent was mixed with the coupling agent at the same mass ratio (5% of the coupling agent mass). Then, resin and filler were added to prepare the coating. Three samples were made for each crosslinking agent. Deionized water droplets were dropped at five different positions on the sample surface using a contact angle meter. The contact angles on the left and right sides of the droplets were measured each time, and the average value of the five positions was recorded as the hydrophobic angle value (in degrees) of the sample. The indentation depth was measured at three points on the sample surface using a Shore hardness tester and converted into Shore D hardness value. The average value of the three points was taken as the hardness value of the sample.
[0073] The maximum values of the hydrophobic angle and hardness for all samples were calculated. Minimum value and maximum value Minimum value The hydrophobic angle value for each sample Perform normalization: Regarding the hardness value Perform normalization: The hydrophobic angle weight was set to 0.6 and the hardness weight to 0.4. These weights were set based on the pre-defined criterion that hydrophobic properties are more important than mechanical properties in coating application scenarios, and the overall score was calculated. ;
[0074] The overall scores of all crosslinking agents are compared, and the crosslinking agent number corresponding to the highest score is selected and recorded as the best crosslinking agent type. This type is determined based on the characteristics of the crosslinking network (e.g., a crosslinking density value exceeding 0.5 mol / L is considered high crosslinking density), curing reaction characteristics (e.g., a curing temperature below 100 degrees Celsius is considered low-temperature curing), and durability performance level (a gloss loss rate of less than 10% after 1000 hours of weathering test is considered high-grade).
[0075] The above scheme eliminates dimensional differences through normalization and sets weights according to application requirements, making the comprehensive evaluation of hydrophobicity and hardness more in line with actual use requirements. It avoids the one-sidedness of optimizing a single index and quickly screens out the best type with balanced performance among a variety of crosslinking agents.
[0076] For example, S4 specifically includes:
[0077] Based on the optimal coupling agent type and concentration as fixed parameters, and using the optimal crosslinking agent type as the variable benchmark, the crosslinking agent dosage is set as the independent variable. Three incremental dosage levels are set (denoted as low level, medium level, and high level, respectively, with specific values determined based on preliminary experimental results or literature recommendations; for example, low level corresponds to lower crosslinking density, and high level corresponds to higher crosslinking density; the horizontal spacing must meet the spacing requirements of the axial point and the center point in the central composite design). According to the principle of central composite design, these three levels are expanded into an experimental scheme that includes center point repetition and axial point, generating a dosage level table.
[0078] Based on the dosage level table, a coating slurry corresponding to each experimental point was prepared, and one sample was prepared for each experimental point. The resin, coupling agent, filler and other components of all samples were kept consistent, and only the actual amount of crosslinking agent added was changed (obtained by reversing the coding level to the actual dosage percentage). The slurry was uniformly coated on a standard substrate (such as tinplate or aluminum plate, the surface of which was cleaned and sanded) using a standard coating process, and cured until completely dry under the set temperature (such as room temperature 25 degrees Celsius) and humidity (50%) conditions. The adhesion grade of each sample was measured point by point using an adhesion tester (such as a pull-out type or a cross-cut type) according to the standard method (such as ISO 2409 or ASTM D3359). The adhesion grade (such as grade 0 to grade 5) was converted into a continuous value (such as grade 0 corresponding to the highest adhesion value of 10 points, grade 5 corresponding to 0 points, and linear interpolation in between). The adhesion readings corresponding to each dosage level were recorded and compiled into an adhesion data matrix.
[0079] Based on the adhesion data matrix, a quadratic polynomial is fitted using the least squares method: ,in This represents the percentage of crosslinking agent used. To obtain an adhesion score, the normal equation is solved. Obtaining regression coefficients (in the equation) To design a matrix containing constant terms, linear terms, and quadratic terms, (where the adhesion score vector is used); take the first derivative of the regression equation. ,make Solve for the extreme points ,like If the dosage is within the range of 1% to 5%, the optimal crosslinking agent dosage is determined directly. If it exceeds the range, the boundary value is taken. The mass addition ratio corresponding to this dosage is the optimal dosage percentage. The curing density level is calculated by the molar mass of the crosslinking agent (if the molecular weight of the crosslinking agent is 200 grams per mole, then the curing density corresponding to a dosage percentage of 3% is 0.15 moles per kilogram of resin). The formulation compatibility coefficient is expressed by calculating the ratio of the optimal dosage to the resin mass.
[0080] The above scheme uses response surface methodology to construct a quadratic regression model and derive it analytically. It accurately fits the functional relationship between adhesion and crosslinking agent dosage with the fewest experimental points. The extreme value solution directly outputs the optimal dosage, realizing a precise upgrade from qualitative screening to quantitative optimization.
[0081] For example, S5 specifically includes:
[0082] The filler mass is set to a fixed proportion of the resin mass. This fixed proportion is derived from the filler proportion value corresponding to the minimum adhesion difference of the optimal particle size combination. 100 parts of resin mass are multiplied by the above fixed proportion to generate the filler mass parts. The multiplication operation is expressed as: filler mass parts = 100 × fixed proportion, where the fixed proportion is a dimensionless ratio.
[0083] Based on the filler mass parts, the optimal coupling agent concentration is divided by the filler mass parts, and then multiplied by 100 to obtain the relative filler mass percentage. The calculation formula is: Relative filler percentage = (Optimal coupling agent concentration / Filler mass parts) × 100. Multiplying this percentage by the filler mass parts gives the coupling agent mass parts, i.e., Coupling agent mass parts = Filler mass parts × (Relative filler percentage / 100). Simultaneously, the optimal crosslinking agent dosage (obtained from the percentage value in step S4 by taking the extreme value of the derivative of the regression equation) is divided by the resin mass reference of 100 parts, and then multiplied by 100 to obtain the relative resin mass percentage. The calculation formula is: Relative resin percentage = (Optimal crosslinking agent dosage / 100) × 100. Multiplying this percentage by the resin mass reference of 100 parts gives the crosslinking agent mass parts, i.e., Crosslinking agent mass parts = 100 × (Relative resin percentage / 100).
[0084] Using the calculated filler, coupling agent, and crosslinking agent mass parts, and taking 100 parts of resin as a baseline, the three are uniformly calculated according to their corresponding baseline quantities. The specific calculation process is as follows: divide the filler mass parts by 100 parts of resin to obtain the mass ratio of filler to resin; divide the coupling agent mass parts by 100 parts of resin to obtain the mass ratio of coupling agent to resin; divide the crosslinking agent mass parts by 100 parts of resin to obtain the mass ratio of crosslinking agent to resin. These three ratios are the proportion coefficients of each component under the 100 parts of resin baseline. Then, taking 100 parts of resin as a baseline, multiply the filler ratio by 100 to obtain the filler parts, the coupling agent ratio by 100 to obtain the coupling agent parts, and the crosslinking agent ratio by 100 to obtain the crosslinking agent parts. Combine these proportional relationships to generate the formulation component quantity values, which are represented by a three-element vector. According to the formula component values, the corresponding mass of filler, coupling agent, crosslinking agent and other raw materials (such as dispersant, defoamer, leveling agent, film-forming aid, etc., the proportions of which are set according to conventional ratios) are added to the mixing tank in sequence, mixed evenly under high-speed dispersion conditions, and then transferred to a grinding mill for grinding to obtain the hydrophobic environmentally friendly coating product.
[0085] The above scheme unifies all the optimal parameters selected in the early stage to the resin standard through precise proportional conversion, forming a quantifiable and reproducible formula, and completes the preparation through standardized dispersion and grinding processes, ensuring the effective transformation of laboratory optimization results into actual production. Example 2
[0086] This embodiment discloses a hydrophobic and environmentally friendly coating, which is prepared using the preparation method described in Example 1.
[0087] The above description only illustrates the preferred embodiments of the present invention and should not be construed as limiting the scope of the claims. The present invention is not limited to the above embodiments, and variations in its specific structure are permitted. All modifications made within the scope of the independent claims of this invention are also within the scope of protection of this invention.
Claims
1. A method for preparing a hydrophobic and environmentally friendly coating, characterized in that, The preparation method includes the following steps: S1. The bonding strength between each silane coupling agent and the filler is obtained by dynamic mechanical analysis, and the bonding strength values are compared to screen out the best coupling agent type. Then, a stratified gradient concentration experiment is carried out to obtain bonding force data, and the best coupling agent concentration is screened out by comparison. S2. Based on the optimal coupling agent type and optimal coupling agent concentration as fixed parameters, an orthogonal experimental design was used to arrange experiments on particle size combinations of micron-sized and nano-sized fillers to obtain adhesion data for each combination. Range analysis was performed to compare and select the optimal particle size combination. S3. Based on the optimal coupling agent type and optimal coupling agent concentration as fixed parameters, multiple crosslinking agents are selected to prepare coatings in the same proportion. After detecting the hydrophobic angle value and hardness value, the values are normalized and then the weighted sum is calculated. The weighted sum values are compared to screen out the optimal crosslinking agent type. S4. Based on the optimal type of crosslinking agent as the variable benchmark, and with the optimal coupling agent type and optimal coupling agent concentration as fixed parameters, a quadratic regression equation between adhesion and crosslinking agent dosage is constructed using the response surface methodology. The dosage value corresponding to the extreme point of the quadratic regression equation is calculated by taking the derivative of the quadratic regression equation as the optimal crosslinking agent dosage. S5. Preparation of hydrophobic environmentally friendly coatings based on optimal coupling agent concentration, optimal particle size combination, optimal crosslinking agent type, and optimal crosslinking agent dosage.
2. The method for preparing a hydrophobic environmentally friendly coating according to claim 1, characterized in that, In step S1, the bonding strength between each silane coupling agent and the filler is obtained through dynamic mechanical analysis, and the bonding strength values are compared to screen out the optimal coupling agent type. Specifically, this includes: Data on the molecular weight, polarity, mixing ratio, temperature, and humidity characteristics of candidate silane coupling agents, as well as data on filler particle size and surface energy characteristics, are obtained. These data are then input into a Gaussian process regression model for training to output a dynamic simulation model. Three dynamic mechanical analyses were performed on each candidate silane coupling agent and filler under conditions of 25 degrees Celsius and 50% humidity. The measured binding strength data of each experiment were collected and the average value of the three experiments was calculated as the average binding strength. The average binding strength, along with the coupling agent identification and experimental conditions, was packaged into a measured feedback dataset.
3. The method for preparing a hydrophobic environmentally friendly coating according to claim 2, characterized in that, Also includes: The dynamic simulation model and the measured feedback dataset are called. The hyperparameters of the dynamic simulation model are iteratively corrected using the Bayesian update algorithm. The average binding strength in the measured feedback dataset is used as the observation value to update the posterior distribution of the prediction function, and the corrected dynamic simulation model is output. The characteristic data of all candidate silane coupling agents are input into the modified dynamic simulation model for prediction. The predicted binding strength value of each coupling agent is obtained. The coupling agent type corresponding to the first position in the ranking is selected as the best coupling agent type. Based on the optimal coupling agent type, five gradient concentration levels of 0.5%, 1.0%, 1.5%, 2.0%, and 2.5% were designed as the percentage of filler mass. For each group of three samples, the binding force was measured and the average value was calculated. The modified dynamic simulation model was used to predict the binding force. The measured average value and the predicted value were weighted and summed with weights of 0.6 and 0.4, respectively. The concentration corresponding to the maximum weighted sum was selected as the optimal coupling agent concentration.
4. The method for preparing a hydrophobic environmentally friendly coating according to claim 1, characterized in that, The step of performing range analysis and comparison in S2 to screen out the optimal particle size combination includes the following steps: By collecting adhesion data, performing mean normalization, calculating the difference between each data point and the mean and dividing it by the standard deviation, standardized adhesion data is obtained. Pearson correlation coefficients between each parameter and the standardized adhesion data are calculated, and the top four are selected as dominant features based on the magnitude of the correlation coefficients. Based on the dominant features and standardized adhesion data, principal component analysis is used to calculate the covariance matrix. The top two principal components with eigenvalues accounting for more than 85% are selected to construct a dimension-reduced feature matrix. The dimension-reduced feature matrix and standardized adhesion data are input into a self-organizing map network. The winning node is determined by Euclidean distance calculation and the weights are adjusted. When the weight update magnitude exceeds twice the standard deviation, an update is performed. The group label vector and the dynamically adjusted standardized adhesion data are output.
5. The method for preparing a hydrophobic environmentally friendly coating according to claim 4, characterized in that, Also includes: For the dynamically adjusted standardized adhesion data, the sparse data part is identified based on the sparsity threshold. The corresponding group label vector is input into the generative adversarial network. The generator uses a multilayer perceptron to generate the complete data. The discriminator calculates the error based on the cross-entropy loss function. When the loss value is lower than 0.01, the completed adhesion data is output. The completed adhesion data and the dynamically adjusted standardized adhesion data are merged into a comprehensive adhesion dataset. Based on the comprehensive adhesion dataset, the average value and range of the adhesion data corresponding to each particle size combination are calculated to obtain the set of range values; Based on the range set, each particle size combination is sorted in ascending order according to the range value, and the combination with the smallest range value is selected as the optimal particle size combination.
6. The method for preparing a hydrophobic environmentally friendly coating according to claim 1, characterized in that, S3 specifically includes: Using the optimal coupling agent type and optimal coupling agent concentration as fixed parameters, several crosslinking agents were selected and mixed with the coupling agent in the same proportion to prepare coatings. The hydrophobic angle values of each coating were recorded using a contact angle measuring instrument, and the hardness values of each coating were recorded using a hardness tester to generate a hydrophobic angle hardness dataset. Based on the hydrophobic angle and hardness dataset, the hydrophobic angle value and hardness value are mapped to intervals using the maximum and minimum value normalization method. For each crosslinking agent, the normalized hydrophobic angle value and hardness value are weighted and summed according to preset weights to obtain the comprehensive performance score set. For the comprehensive performance score set, compare the scores of all crosslinking agents, select the crosslinking agent with the highest score, record the crosslinking agent number, and obtain the optimal crosslinking agent type.
7. The method for preparing a hydrophobic environmentally friendly coating according to claim 1, characterized in that, S4 specifically includes: The optimal type of crosslinking agent was used as the baseline variable, the amount of crosslinking agent was set as the independent variable, three incremental amount levels were set, the experimental points were arranged according to the central composite design principle, and an amount level table was generated. Based on the dosage level table, the coating slurry was prepared and applied to the standard substrate. After curing, the adhesion tester was used to measure the adhesion point by point according to the standard method. The adhesion readings corresponding to each dosage level were recorded and compiled into an adhesion data matrix. Based on the adhesion data matrix, a quadratic polynomial equation is fitted using the least squares method to obtain the regression coefficient of adhesion with respect to the amount of crosslinking agent. The first derivative of the regression equation is calculated, and the derivative is set to zero to solve for the amount of crosslinking agent at the extreme point, thus obtaining the optimal amount of crosslinking agent.
8. The method for preparing a hydrophobic environmentally friendly coating according to claim 1, characterized in that, S5 specifically includes: The resin mass baseline is set to 100 parts. The filler mass corresponding to the optimal particle size combination is set as a fixed proportion of the resin mass. The filler mass parts are generated by multiplying the resin mass baseline by this proportion. Based on the filler mass fraction, the optimal coupling agent concentration value is divided by the filler mass fraction and then multiplied by 100 to obtain the relative filler mass percentage. This percentage is then multiplied by the filler mass fraction. Simultaneously, the optimal crosslinking agent dosage value is divided by the resin mass reference and then multiplied by 100 to obtain the relative resin mass percentage. This percentage is then multiplied by the resin mass reference to obtain the coupling agent mass fraction and crosslinking agent mass fraction. The mass fractions of filler, coupling agent, and crosslinking agent are called up. Based on 100 parts of resin, the three are standardized according to their corresponding reference amounts. The proportions are combined to obtain the respective part numbers and generate the component values of the formulation.
9. The method for preparing a hydrophobic environmentally friendly coating according to claim 1, characterized in that, S5 further includes adding filler, coupling agent, crosslinking agent and other raw materials to a mixing tank in sequence according to the formula component values, dispersing them at high speed, and then transferring them to a grinding mill for grinding, thereby obtaining the hydrophobic environmentally friendly coating.
10. A hydrophobic and environmentally friendly coating, characterized in that: It is prepared by the preparation method described in any one of claims 1-9.
