Automatic control system and method for preparing water-based zinc-aluminum protective coating for metal building materials

Abstract

The present invention discloses an automatic control system and method for preparing water-based zinc-aluminum protective coatings for metal building materials, and relates to the field of coating preparation. The method includes adhesive preparation; aluminum powder modification treatment; coating liquid preparation, including optimizing and selecting control parameters in coating liquid preparation based on an optimized radial neural network model; coating liquid performance detection, and obtaining coating, drying, and curing parameters based on the coating liquid performance based on an improved SVM algorithm; and testing the performance of the coating. The present invention realizes the control of coating preparation process parameters based on an intelligent algorithm, and adjusts the coating, drying, and curing parameters according to the performance of the coating liquid, and realizes the selection of appropriate process parameters to achieve better coating performance.

Description

Technical Field

[0001] The invention relates to the field of coating preparation, in particular to an automatic control system and method for preparing a water-based zinc-aluminum protective coating for metal building materials. Background Art

[0002] Metal building materials are widely used in various aspects of buildings due to their excellent durability, decorativeness and processability. However, metals are very susceptible to corrosion in the atmosphere, which seriously affects the service life of building materials. Among all anti-corrosion technologies and methods, anti-corrosion coatings are the most effective, economical and commonly used methods, and in some areas, coatings are almost the only feasible anti-corrosion method. Among them, water-based zinc-aluminum coatings have long been widely used in metal building materials due to their excellent rust resistance and durability.

[0003] In the prior art, the preparation process of water-based zinc-aluminum protective coatings usually uses experience to determine the preparation parameters, and the coating coating process is relatively fixed, without considering the different characteristics of the coating during coating, resulting in large differences in the final performance of the coating and unstable performance. Summary of the invention

[0004] In order to solve the above technical problems, the present invention provides an automatic control method and system for preparing a water-based zinc-aluminum protective coating for metal building materials.

[0005] In order to solve the above-mentioned technical problems and achieve the purpose of the invention, the present invention is implemented by the following technical solutions:

[0006] An automatic control method for preparing a water-based zinc-aluminum protective coating for metal building materials, comprising:

[0007] Adhesive preparation, including preparation of high modulus organic hybrid silica sol adhesive;

[0008] Aluminum powder modification treatment: select the method of organic silicon modification treatment to modify the surface of pigment aluminum powder;

[0009] Coating preparation, including optimizing the ratio of binder to zinc and aluminum powders, stirring speed, and standing time in coating preparation based on an optimized radial neural network model;

[0010] Detect coating liquid performance and obtain coating, drying and curing parameters based on the improved SVM algorithm according to the coating liquid performance;

[0011] The coating properties are tested, including coating thickness, coating amount, adhesion, appearance, and corrosion resistance.

[0012] Furthermore, the preparation of the high modulus organic hybrid silica sol adhesive comprises:

[0013] Adding a low modulus lithium silicate solution into a preparation container, stirring with a magnetic stirrer at a first stirring speed for a first stirring time, and heating to a first heating temperature;

[0014] Adding a high modulus silica sol solution and deionized water at a first dropping speed and a second dropping speed respectively, adjusting the stirring speed to a second stirring speed and stirring for a second stirring time, and adjusting the heating temperature to a second heating temperature;

[0015] After the high modulus silica sol solution and deionized water are dripped, stirring is continued at a third stirring speed for a third stirring time, and the temperature is adjusted to a third heating temperature to fully dissolve the solution;

[0016] Add the silane coupling agent dropwise at a third dropping speed, adjust the temperature to a fourth heating temperature, adjust the stirring speed to a fourth stirring speed, and adjust the stirring time to a fourth stirring time;

[0017] After the reaction is completed, the solution is transferred to a sealed container and sealed, and cooled to room temperature to obtain a high modulus mixed silica sol.

[0018] Furthermore, before adding the low modulus lithium silicate solution to the preparation container, the method also includes selecting control parameters for magnetic stirring based on a decision tree algorithm, wherein the control parameters include solution particle size adjustment stirring speed, stirring time, high modulus silica sol, deionized water, silane coupling agent dripping speed and heating temperature.

[0019] Furthermore, the selection of control parameters for magnetic stirring based on the decision tree algorithm includes:

[0020] The particle size of the stirred solution was measured using a laser particle size analyzer;

[0021] Establish a decision tree model and conduct model training;

[0022] Calculate the output parameters;

[0023] Input setting parameters into the decision tree model, including preset solution particle size, mass of solution to be prepared, and mass of low modulus lithium silicate solution; obtain model output values, which include a first stirring speed, a second stirring speed, a third stirring speed, a fourth stirring speed, a first dripping speed, a second dripping speed, a third dripping speed, a first stirring time, a second stirring time, a third stirring time, a fourth stirring time, and a first heating temperature, a second heating temperature, a third heating temperature, and a fourth heating temperature.

[0024] Furthermore, the establishment of a decision tree model and model training include:

[0025] Given training set data , loss function ;in, is the sample in the training set, is the calculated value;

[0026] Initialize the weak learner: ;in, is a decision tree with only one root node. is the loss function, N is the number of training samples, is the i-th sample in the training sample set; s is the constant value that minimizes the loss function;

[0027] Compute the residual:

[0028] The residual for each training sample is as follows: Where m represents the number of iterations, m=1,2,…,Max, is the calculated value;

[0029] The calculated residual value is used as the true value of the next training. , i=1,2,…,N as the sample data for the next training, and a tree with j leaf nodes distributed in Decision tree of

[0030] Calculate the best fit value for the leaf region , where j=1,2,…,J; ;in, is the calculated value of the m-1th iteration, c is the constant value of minimizing the loss function, The value of the function parameter to obtain the minimum value;

[0031] Update the strong learner: ; Where I is the indicator function: Accumulate to get the final decision tree: Among them, Max is the maximum number of iterations, and J is the total number of leaf nodes.

[0032] Furthermore, the optimized radial neural network model based on which the ratio of the binder to the metal powder zinc and aluminum powder, the stirring speed, and the standing time in the preparation of the coating liquid are optimized also includes:

[0033] The optimized radial neural network model uses an improved Gaussian function to construct the kernel function, and the calculation equation is: in, is a variable, Calculate the target value for the kernel function, Calculate the error for the model.

[0034] Furthermore, the coating, drying and curing parameters obtained based on the improved SVM algorithm according to the coating liquid performance also include:

[0035] The viscosity, fineness, stability, and thermal properties of the coating liquid were tested as inputs to the SVM model;

[0036] Build the SVM model;

[0037] The classification decision function of the nonlinear support vector machine is set as: Where M is the number of input data, is the optimal solution of the Lagrange multiplier, is the class marker, is the input data independent variable, is the kernel function, is the optimal segmentation hyperplane parameter, ,in is the width parameter of the function;

[0038] Input the coating liquid viscosity, fineness, stability, and thermal properties, and output process parameters according to the SVM model for coating, drying, and curing; the output process parameters include the number of coatings, drying temperature, drying time, and curing temperature and curing time.

[0039] Furthermore, the detection of the viscosity, fineness, stability and thermal properties of the coating liquid includes: detecting the viscosity of the coating liquid using a viscometer; detecting the fineness of the coating liquid based on a laser diffraction method; detecting the appearance of the coating liquid based on image data obtained by a camera, and determining the stability of the coating liquid based on bubbling and agglomeration; and obtaining the thermal properties of the coating liquid based on a gravimetric-thermal analyzer.

[0040] Furthermore, the SVM model establishment also includes:

[0041] Optimizing the SVM model based on the swarm intelligence algorithm includes the following steps:

[0042] Initialize the population, maximum number of iterations;

[0043] Normalize the data in the dataset and divide the normalized dataset into a training set and a test set;

[0044] Select the objective function;

[0045] Calculate individual fitness values ​​and determine the best individual; each animal performs exploration and development, and determines the next behavior based on the results: exploration and development; after a round of search, store the best value in Xbest based on the fitness value of each animal;

[0046] In the exploration phase, the position of individual animals is updated by wavelet function perturbation combined with Cauchy distribution as follows: ;in, is the updated position, is the position before updating, is the wavelet function perturbation factor, is a random number in [1, 2], r is a random number in [0, 1], represents vector multiplication, is the standard Cauchy distribution;

[0047] During the development phase, a win-win idea is introduced to update the location, as follows: ; where r is a random number in [0, 1], , are the upper and lower limits of the j-th dimension variable updated with the number of iterations, R is the interaction factor between the current individual and the random individual, is the adjustment factor;

[0048] Determine the penalty factor and width parameters that need to be tuned in the SVM model ；

[0049] Determine whether the termination condition is met. During the iterative optimization process, iterate based on the fitness value. When the number of iterations meets the termination condition, terminate the iteration and output the parameters corresponding to the best animal individual position and the optimal solution. Otherwise, return to calculate the individual fitness value, determine the optimal individual step and continue iterating.

[0050] After the optimization is completed, the parameters of the output optimal solution are used as the SVM model parameters to obtain the best classification model after training.

[0051] The present invention also provides an automatic control system for preparing a water-based zinc-aluminum protective coating for metal building materials, which is used to implement the automatic control method for preparing a water-based zinc-aluminum protective coating for metal building materials, comprising:

[0052] Adhesive preparation module, which is used to prepare high modulus organic hybrid silica sol adhesive; it selects control parameters for magnetic stirring based on a decision tree algorithm, including solution particle size adjustment stirring speed, and the dripping speed of high modulus silica sol, deionized water, and silane coupling agent;

[0053] An aluminum powder modification treatment module is used to select a method of organic silicon modification treatment to perform surface modification on the pigment aluminum powder;

[0054] The coating preparation module optimizes the ratio of binder to metal powder zinc and aluminum powder, stirring speed, and standing time in coating preparation based on the optimized radial neural network model;

[0055] A post-processing module, which includes coating liquid performance detection, and obtains coating, drying, and curing parameters based on the coating liquid performance based on an improved SVM algorithm, and performs processing according to the determined parameters;

[0056] The test module is used to test the performance of the coating, including coating thickness, coating amount, adhesion, appearance, and corrosion resistance.

[0057] Compared with the prior art, the present invention has the following beneficial effects:

[0058] The present invention adjusts coating, drying and curing parameters according to the performance of the coating liquid based on an artificial intelligence algorithm, thereby realizing the selection of appropriate process parameters to achieve better coating performance.

[0059] In the preparation of the adhesive, the present invention selects control parameters for magnetic stirring based on a decision tree algorithm, including adjusting the stirring speed according to the solution particle size, and the dripping speed of high modulus silica sol, deionized water, and silane coupling agent, thereby achieving the selection of optimal control parameters, avoiding the subjective judgment of different manual selections in the prior art, and improving the manufacturing quality.

[0060] The present invention optimizes and selects the ratio of the binder to the metal powder zinc and aluminum powder, the stirring speed and the standing time in the coating liquid preparation based on the optimized radial neural network model in the coating liquid preparation. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0062] Figure 1 It is a schematic flow chart of an automatic control method for preparing a water-based zinc-aluminum protective coating for metal building materials according to an embodiment of the present application. DETAILED DESCRIPTION

[0063] The embodiments of the present disclosure are described in detail below with reference to the accompanying drawings.

[0064] The following describes the embodiments of the present disclosure through specific examples, and those skilled in the art can easily understand other advantages and effects of the present disclosure from the contents disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present disclosure, rather than all of the embodiments. The present disclosure can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present disclosure. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present disclosure.

[0065] It should also be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present disclosure. The illustrations only show components related to the present disclosure rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the form, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0066] See also Figure 1 , an automatic control method for preparing a water-based zinc-aluminum protective coating for metal building materials comprises the following steps:

[0067] S1: Adhesive preparation;

[0068] The adhesive uses low modulus lithium silicate solution as the basic raw material. By adding silica sol to the low modulus lithium silicate solution, the silicon-oxygen molar ratio of the lithium silicate solution is increased. Finally, a high modulus organic mixed silica sol adhesive is prepared as a binder to replace traditional chromic acid. The specific preparation process includes the following steps:

[0069] S11: adding a low modulus lithium silicate solution into a preparation container, stirring with a magnetic stirrer at a first stirring speed for a first stirring time, and heating to a first heating temperature;

[0070] S12: dripping the high modulus silica sol solution and deionized water at a first dripping speed and a second dripping speed respectively, adjusting the stirring speed to a second stirring speed and stirring for a second stirring time, and adjusting the heating temperature to a second heating temperature;

[0071] S13: After the high modulus silica sol solution and deionized water are dripped, stirring is continued at a third stirring speed for a third stirring time, and the temperature is adjusted to a third heating temperature to fully dissolve the solution;

[0072] S14: dripping the silane coupling agent at a third dripping speed, adjusting the temperature to a fourth heating temperature, the stirring speed to a fourth stirring speed, and the stirring time to a fourth stirring time;

[0073] S15: After the reaction is completed, the solution is transferred to a sealed container and sealed, and cooled to room temperature to obtain a high modulus mixed silica sol.

[0074] In the prior art, the magnetic stirring time, stirring speed, and solution dripping speed are generally determined by experience. The empirical values ​​used by different technicians vary greatly, which also causes large differences in the final stirring results and makes it impossible to achieve an adhesive with high transparency and stability.

[0075] Therefore, the present invention selects control parameters for magnetic stirring based on a decision tree algorithm, wherein the control parameters include the stirring speed for adjusting the solution particle size, the dripping speed of high modulus silica sol, deionized water, and silane coupling agent, and comprises the following steps:

[0076] S101: using a laser particle size analyzer to detect the particle size of the stirred solution;

[0077] S102: Establish a decision tree model and perform model training;

[0078] S1021: Given training set data , loss function ;in, is the sample in the training set, is the calculated value;

[0079] Initialize the weak learner: ;in, is a decision tree with only one root node. is the loss function, N is the number of training samples, is the i-th sample in the training sample set; s is the constant value that minimizes the loss function.

[0080] S1022: Calculate residuals:

[0081] The residual for each training sample is as follows: Where m represents the number of iterations, m=1,2,…,Max, is the calculated value.

[0082] S1023: The calculated residual value is used as the true value of the next training. , i=1,2,…,N as the sample data for the next training, and a tree with j leaf nodes distributed in Decision tree of

[0083] S1024: Calculate the best fit value of the leaf area , where j=1,2,…,J; in, is the calculated value of the m-1th iteration, c is the constant value of minimizing the loss function, and c is the constant value of minimizing the loss function. The value of the function parameter that obtains the minimum value.

[0084] S1025: Update the strong learner: ; Where I is the indicator function: S1026: Accumulate and obtain the final decision tree: Among them, Max is the maximum number of iterations, and J is the total number of leaf nodes.

[0085] S103: Calculate and obtain output parameters;

[0086] Input setting parameters into the decision tree model, including preset solution particle size, mass of solution to be prepared, and mass of low modulus lithium silicate solution; obtain model output values, which include a first stirring speed, a second stirring speed, a third stirring speed, and a fourth stirring speed, dripping speeds of high modulus silica sol, deionized water, and silane coupling agent (first dripping speed, second dripping speed, and third dripping speed), a first stirring time, a second stirring time, a third stirring time, and a fourth stirring time, and a first heating temperature, a second heating temperature, a third heating temperature, and a fourth heating temperature.

[0087] S2: aluminum powder modification treatment;

[0088] Selecting the method of organic silicon modification treatment to modify the surface of the pigment aluminum powder includes the following steps:

[0089] S21: Weigh a certain amount of flaky aluminum powder and disperse it into a certain volume of isopropanol, and perform ultrasonic dispersion;

[0090] S22: Pour the dispersed sample into a container and heat and stir at a constant temperature;

[0091] S23: Weigh a certain amount of silane coupling agent and add it into a container, and drop a certain amount of ammonia water and distilled water into it;

[0092] S24: After the reaction is completed, the aluminum powder is washed with anhydrous ethanol and further dried to obtain the modified aluminum powder.

[0093] S3: coating solution preparation;

[0094] First, the prepared binder is mixed with the pigment metal powders zinc and aluminum in a certain mass ratio to prepare the coating; during the preparation process, the metal powder is slowly added to the binder and fully mechanically stirred; secondly, after the pigment aluminum powder is added, the auxiliary agent is added and continued to be stirred to allow it to be fully dispersed; finally, the coating is allowed to stand for use.

[0095] In the preparation of the coating liquid, the ratio of the binder to the metal powder zinc and aluminum powder, the stirring speed, and the standing time comprehensively affect the various performance indicators of the coating liquid. In the prior art, when performing this step, the ratio, stirring speed, and standing time are roughly determined based on experience, and the coating liquid performance cannot be optimized. Therefore, in order to obtain the optimal coating liquid preparation parameters, the present invention optimizes the ratio of the binder to the metal powder zinc and aluminum powder, the stirring speed, and the standing time in the coating liquid preparation based on the optimized radial neural network model, which specifically includes the following steps:

[0096] The optimized neural network model uses an improved Gaussian function to construct a kernel function, and the calculation equation is: in, is a variable, Calculate the target value for the kernel function, Calculate the error for the model.

[0097] The improved model uses the gradient descent method to optimize its calculation in the process of seeking the optimal solution.

[0098] Aiming at the problem of slow convergence rate of traditional radial basis function neural network model, the gradient optimization function is introduced to optimize the calculation of its variables, which can improve the convergence rate of traditional radial basis function neural network model and thus improve its prediction accuracy in nonlinear changing data series.

[0099] S4: coating liquid performance detection, and based on the coating liquid performance based on the improved SVM algorithm to obtain the optimal parameters of coating, drying and curing;

[0100] In the prior art, the selection of coating, drying and curing process parameters is usually determined based on experience, and the performance of the coating liquid is usually rarely considered, and it is generally only judged by manual observation with the naked eye. The performance of the coating liquid has a great influence on the subsequent process. Selecting appropriate process parameters for coating, drying and curing according to coating liquids with different performances has a significant improvement on the final performance of the coating. Therefore, the present invention selects appropriate process parameters according to the performance of the coating liquid based on the improved SVM algorithm to achieve better coating performance. Specifically, it includes the following steps:

[0101] S41: Testing the viscosity, fineness, stability and thermal properties of the coating liquid as input to the SVM model;

[0102] Specifically, the viscosity of the coating liquid is tested using a viscometer; the fineness of the coating liquid is tested based on the laser diffraction method; the appearance of the coating liquid is tested based on image data obtained by a camera, and the stability of the coating liquid is determined based on bubbling and agglomeration conditions; and the thermal properties of the coating liquid are obtained based on a gravimetric-thermal analyzer.

[0103] S42: Establishing SVM model;

[0104] The classification decision function of the nonlinear support vector machine is set as: Where M is the number of input data, is the optimal solution of the Lagrange multiplier, is the class marker, is the input data independent variable, is the kernel function, is the optimal segmentation hyperplane parameter, ,in is the width parameter of the function;

[0105] The optimal segmentation hyperplane of the SVM model is related to the penalty factor and the width parameter, so it is necessary to optimize the parameters of the two to improve the classification and diagnosis performance of the SVM.

[0106] The present invention optimizes the SVM model based on a swarm intelligence algorithm, comprising the following steps:

[0107] S421: Initialize the population, maximum number of iterations;

[0108] S422: normalizing the data in the data set, and dividing the normalized data set into a training set and a test set;

[0109] S423: Selecting an objective function;

[0110] S424: Calculate individual fitness values ​​and determine the best individual. Each animal performs exploration and development, and the next behavior is determined based on the results: exploration and development. After a round of search, the best value is stored in Xbest based on the fitness value of each animal.

[0111] In the exploration phase, the position of individual animals is updated using wavelet function perturbations as follows: ;in, is the updated position, is the position before updating, is the wavelet function perturbation factor, is a random number in [1, 2], r is a random number in [0, 1], represents vector multiplication, is the standard Cauchy distribution.

[0112] In the development stage, the win-win idea is introduced to update the position as follows: ; where r is a random number in [0, 1], , are the upper and lower limits of the j-th dimension variable updated with the number of iterations, R is the interaction factor between the current individual and the random individual, is the adjustment factor.

[0113] By adding the guidance of interactive relationships, the information exchange between the current individual and other individuals is richer, so that the animal individuals will move closer to the global optimal position based on the interactive information, avoiding blind search and improving the overall solution accuracy of the algorithm.

[0114] S425: Determine the penalty factor and width parameter that need to be tuned in the SVM model ；

[0115] S426: Determine whether the termination condition is met. During the iterative optimization process, the iteration is performed based on the fitness value. When the number of iterations meets the termination condition, the iteration is terminated and the parameters corresponding to the best animal individual position are output, and the optimal solution is output. Otherwise, return to step S424.

[0116] S427: The optimization is completed, and the parameters output in step S426 are used as SVM model parameters to obtain the best classification model after training.

[0117] S43: Input the coating liquid viscosity, fineness, stability, and thermal properties, and output process parameters according to the SVM model for coating, drying, and curing; the output process parameters include the number of coatings, drying temperature, drying time, and curing temperature and curing time.

[0118] S5: Testing;

[0119] The coating properties are tested, including coating thickness, coating amount, adhesion, appearance, and corrosion resistance.

[0120] In this embodiment, the coating, drying, and curing parameters are adjusted according to the performance of the coating liquid based on the artificial intelligence algorithm, and the appropriate process parameters are selected to achieve better coating performance. In the production of adhesives, the magnetic stirring control parameters are selected based on the decision tree algorithm, including the stirring speed adjusted by the solution particle size, the dripping speed of high modulus silica sol, deionized water, and silane coupling agent, to achieve the selection of optimal control parameters, avoid the subjective judgment of different manual selections in the prior art, and improve the manufacturing quality. In the preparation of the coating liquid, the ratio of the binder to the metal powder zinc and aluminum powder, the stirring speed, and the standing time in the coating liquid preparation are optimized based on the optimized radial neural network model.

[0121] The embodiment of the present invention also provides an automatic control system for preparing a water-based zinc-aluminum protective coating for metal building materials, comprising:

[0122] Adhesive preparation module, which is used to prepare high modulus organic hybrid silica sol adhesive; it selects control parameters for magnetic stirring based on a decision tree algorithm, including solution particle size adjustment stirring speed, and the dripping speed of high modulus silica sol, deionized water, and silane coupling agent;

[0123] An aluminum powder modification treatment module is used to select a method of organic silicon modification treatment to perform surface modification on the pigment aluminum powder;

[0124] The coating liquid preparation module is used to mix the prepared binder with the pigment metal powder zinc and aluminum in a certain mass ratio. During the preparation process, the metal powder is first slowly added to the binder and fully mechanically stirred. After the pigment aluminum powder is added, the additive is added and continued to be stirred to allow it to be fully dispersed, and finally allowed to stand for coating;

[0125] A post-processing module, which includes coating liquid performance detection, and obtains the optimal parameters of coating, drying and curing based on the coating liquid performance based on the improved SVM algorithm, and performs processing according to the determined parameters;

[0126] The test module is used to test the performance of the coating, including coating thickness, coating amount, adhesion, appearance, and corrosion resistance.

[0127] The embodiments described above are only descriptions of the preferred implementation modes of the present invention, and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. An automatic control method for preparing a water-based zinc-aluminum protective coating for metal building materials, characterized in that: include: Adhesive preparation, including preparation of high modulus organic hybrid silica sol adhesive; Aluminum powder modification treatment: select the method of organic silicon modification treatment to modify the surface of pigment aluminum powder; Coating preparation, including optimizing the ratio of binder to zinc and aluminum powders, stirring speed, and standing time in coating preparation based on an optimized radial neural network model; The optimized radial neural network model uses an improved Gaussian function to construct a kernel function, and the calculation equation is: in, is a variable, Calculate the target value for the kernel function, Calculate the error for the model; Detect coating liquid performance and obtain coating, drying and curing parameters based on the improved SVM algorithm according to the coating liquid performance; The coating, drying and curing parameters obtained based on the improved SVM algorithm according to the coating liquid performance also include: The viscosity, fineness, stability, and thermal properties of the coating liquid were tested as inputs to the SVM model; Build the SVM model; The classification decision function of the nonlinear support vector machine is set as: Where M is the number of input data, is the optimal solution of the Lagrange multiplier, is the class marker, is the input data independent variable, is the kernel function, is the optimal segmentation hyperplane parameter, ,in is the width parameter of the function; The output process parameters include coating times, drying temperature, drying time, curing temperature and curing time; The SVM model establishment also includes: Optimizing the SVM model based on the swarm intelligence algorithm includes the following steps: Initialize the population, maximum number of iterations; Normalize the data and divide the normalized data set into a training set and a test set; Select the objective function; Calculate individual fitness values ​​and determine the best individual; each animal performs exploration and development, and determines the next behavior based on the results: exploration and development; after a round of search, store the best value in Xbest based on the fitness value of each animal; In the exploration phase, the position of individual animals is updated by wavelet function perturbation combined with Cauchy distribution as follows: ;in, is the updated position, is the position before updating, is the wavelet function perturbation factor, is a random number in [1, 2], r is a random number in [0, 1], represents vector multiplication, is the standard Cauchy distribution; During the development phase, a win-win idea is introduced to update the location, as follows: ; where r is a random number in [0, 1], , are the upper and lower limits of the j-th dimension variable updated with the number of iterations, R is the interaction factor between the current individual and the random individual, is the adjustment factor; Determine the penalty factor and width parameters that need to be tuned in the SVM model ； Determine whether the termination condition is met. During the iterative optimization process, iterate based on the fitness value. When the number of iterations meets the termination condition, terminate the iteration and output the parameters corresponding to the best animal individual position and the optimal solution. Otherwise, return to calculate the individual fitness value, determine the optimal individual step and continue iterating. After the optimization is completed, the parameters of the output optimal solution are used as the SVM model parameters to obtain the best classification model after training; The coating properties are tested, including coating thickness, coating amount, adhesion, appearance, and corrosion resistance.

2. The automatic control method for preparing a water-based zinc-aluminum protective coating for metal building materials according to claim 1, characterized in that: The method for preparing the high modulus organic hybrid silica sol adhesive comprises: Adding a low modulus lithium silicate solution into a preparation container, stirring with a magnetic stirrer at a first stirring speed for a first stirring time, and heating to a first heating temperature; Adding a high modulus silica sol solution and deionized water at a first dropping speed and a second dropping speed respectively, adjusting the stirring speed to a second stirring speed and stirring for a second stirring time, and adjusting the heating temperature to a second heating temperature; After the high modulus silica sol solution and deionized water are dripped, stirring is continued at a third stirring speed for a third stirring time, and the temperature is adjusted to a third heating temperature to fully dissolve the solution; Add the silane coupling agent dropwise at a third dropping speed, adjust the temperature to a fourth heating temperature, adjust the stirring speed to a fourth stirring speed, and adjust the stirring time to a fourth stirring time; After the reaction is completed, the solution is transferred to a sealed container and sealed, and cooled to room temperature to obtain a high modulus mixed silica sol.

3. The automatic control method for preparing a water-based zinc-aluminum protective coating for metal building materials according to claim 2, characterized in that: Before adding the low modulus lithium silicate solution into the preparation container, the method further includes selecting control parameters for magnetic stirring based on a decision tree algorithm, wherein the control parameters include stirring speed and stirring time for adjusting the particle size of the solution, dripping speed of high modulus silica sol, deionized water, and silane coupling agent, and heating temperature.

4. The automatic control method for preparing a water-based zinc-aluminum protective coating for metal building materials according to claim 3, characterized in that: The selection of control parameters for magnetic stirring based on the decision tree algorithm includes: The particle size of the stirred solution was measured using a laser particle size analyzer; Establish a decision tree model and conduct model training; Calculate the output parameters; Input setting parameters into the decision tree model, including preset solution particle size, mass of solution to be prepared, and mass of low modulus lithium silicate solution; obtain model output values, which include a first stirring speed, a second stirring speed, a third stirring speed, a fourth stirring speed, a first dripping speed, a second dripping speed, a third dripping speed, a first stirring time, a second stirring time, a third stirring time, a fourth stirring time, and a first heating temperature, a second heating temperature, a third heating temperature, and a fourth heating temperature.

5. The automatic control method for preparing water-based zinc-aluminum protective coating for metal building materials according to claim 4, characterized in that: The step of establishing a decision tree model and performing model training includes: Given training set data , loss function ;in, is the sample in the training set, is the calculated value; Initialize the weak learner: ;in, is a decision tree with only one root node. is the loss function, N is the number of training samples, is the i-th sample in the training sample set; s is the constant value that minimizes the loss function; Compute the residual: The residual for each training sample is as follows: Where m represents the number of iterations, m=1,2,…,Max, is the calculated value; The calculated residual value is used as the true value of the next training. , i=1,2,…,N as the sample data for the next training, and a tree with j leaf nodes distributed in Decision tree of Calculate the best fit value for the leaf region , where j=1,2,…,J; in, is the calculated value of the m-1th iteration, c is the constant value of minimizing the loss function, The value of the function parameter to obtain the minimum value; Update the strong learner: ; Where I is the indicator function: Accumulate to get the final decision tree: Among them, Max is the maximum number of iterations, and J is the total number of leaf nodes.

6. The automatic control method for preparing water-based zinc-aluminum protective coating for metal building materials according to claim 5, characterized in that: The detection of the viscosity, fineness, stability and thermal properties of the coating liquid includes: detecting the viscosity of the coating liquid using a viscometer; detecting the fineness of the coating liquid based on a laser diffraction method; detecting the appearance state of the coating liquid based on image data obtained by a camera, and determining the stability of the coating liquid based on bubbling and agglomeration conditions; and obtaining the thermal properties of the coating liquid based on a gravimetric-thermal analyzer.

7. An automatic control system for preparing a water-based zinc-aluminum protective coating for metal building materials, used to implement the automatic control method for preparing a water-based zinc-aluminum protective coating for metal building materials as described in any one of claims 1 to 6, characterized in that: include: Adhesive preparation module, which is used to prepare high modulus organic hybrid silica sol adhesive; it selects control parameters for magnetic stirring based on a decision tree algorithm, including solution particle size adjustment stirring speed, and the dripping speed of high modulus silica sol, deionized water, and silane coupling agent; An aluminum powder modification treatment module is used to select a method of organic silicon modification treatment to perform surface modification on the pigment aluminum powder; The coating preparation module optimizes the ratio of binder to metal powder zinc and aluminum powder, stirring speed, and standing time in coating preparation based on the optimized radial neural network model; A post-processing module, which includes coating liquid performance detection, and obtains coating, drying, and curing parameters based on the coating liquid performance based on an improved SVM algorithm, and performs processing according to the determined parameters; The test module is used to test the performance of the coating, including coating thickness, coating amount, adhesion, appearance, and corrosion resistance.

Images