Production control method of nano-modified anti-corrosion coating based on genetic algorithm

Abstract

The present invention relates to a production control method for nano-modified anti-corrosion coatings, and specifically to a production control method for nano-modified anti-corrosion coatings based on genetic algorithms. First, a multi-objective optimization fitness function is defined to comprehensively evaluate the dispersion, viscosity, hardness and corrosion resistance under the coating formula, and consider the production cost; then, the component ratio and process parameters in the coating formula are encoded using real number coding, and an initial population is generated; then, the dominant formula is selected through the roulette wheel selection method in genetic operations, and the formula diversity is improved using a single-point or multi-point crossover method, and the component ratio or process parameters are randomly mutated to explore new formulas; finally, an algorithm module is developed to regulate the size, shape and distribution of nanomaterials, and the reaction conditions are adjusted according to the optimization results of the genetic algorithm to achieve the optimization of coating performance.

Description

Technical Field

[0001] The invention relates to a production control method for a nano-modified anti-corrosion coating, in particular to a production control method for a nano-modified anti-corrosion coating based on a genetic algorithm. Background Art

[0002] The current production control methods for nano-modified anti-corrosion coatings have deficiencies and drawbacks in many aspects, which may affect production efficiency, product quality, and environmental and economic sustainability. First, traditional coating production control methods often rely on experience and manual adjustments, lacking a high degree of automation and intelligence. This results in the inability to respond to changes on the production line in real time during the production process, such as batch differences in raw materials, changes in environmental conditions, etc., which affects the quality and consistency of the final product. For example, when the ambient temperature or humidity changes, it is difficult for traditional methods to quickly adjust production parameters to ensure that the performance of the coating is not affected. In addition, this method that relies on manual experience also leads to low production efficiency, because each batch of raw materials needs to be tested and adjusted to determine the best formula and production conditions. Secondly, the existing production control methods are usually iterative and time-consuming processes in formula development and optimization. The R&D team needs to test the performance of different formulas through a large number of experiments, which is not only time-consuming but also costly. Each new test requires a lot of raw materials, energy and manpower, and due to the lack of precise control and prediction tools, many tests may not produce useful results. The inefficiency of this method prolongs the development cycle of new products and makes it difficult to quickly respond to market demand and technological advances. In addition, traditional control methods face challenges in dealing with complex coating systems, especially in terms of the dispersion and stability of nanomaterials and their compatibility with matrix materials. Due to their unique properties, such as high specific surface area and nano-effect, nanomaterials' behavior in coatings is difficult to predict and control with simple models. This leads to problems such as aggregation, sedimentation or reaction with matrix materials in the production process, which may affect the performance and quality of the coating. Existing production control methods often have difficulty in real-time monitoring and adjustment of these complex nanomaterial behaviors, increasing production uncertainty and risk. Economically, due to the low efficiency and high consumption of traditional methods, the production cost is relatively high. This includes not only direct raw material and energy costs, but also indirect costs due to factors such as low production line downtime and high product failure rate. These economic benefits are insufficient to limit the development and expansion capabilities of enterprises in the fierce market competition.

[0003] In summary, the existing production control methods of nano-modified anti-corrosion coatings have obvious deficiencies and disadvantages in terms of automation, intelligence, efficiency, environmental friendliness and economy. Summary of the invention

[0004] The purpose of the present invention is to provide a method for controlling the production of nano-modified anti-corrosion coatings based on genetic algorithms, thereby solving some of the drawbacks and shortcomings pointed out in the background technology.

[0005] The present invention solves the above-mentioned technical problems by adopting the following technical solutions: A method for controlling the production of nano-modified anti-corrosion coatings based on genetic algorithms, comprising: firstly defining a fitness function for multi-objective optimization, comprehensively evaluating the dispersion, viscosity, hardness and corrosion resistance under the coating formula, and considering the production cost;

[0006] Then, the component ratios and process parameters in the coating formula are encoded using real number coding, and the initial population is generated;

[0007] Then, the dominant formula is selected through the roulette wheel selection method in genetic operation, the formula diversity is improved by using single-point or multi-point crossover method, and the ingredient ratio or process parameters are randomly mutated to explore new formulas;

[0008] Finally, an algorithm module was developed to control the size, shape and distribution of nanomaterials, and the reaction conditions were adjusted according to the optimization results of the genetic algorithm to achieve the optimization of coating performance.

[0009] Furthermore, the fitness function construction of the multi-objective optimization includes:

[0010] S1. First, a high-dimensional nonlinear regression model is constructed to capture the effects of individual components and interactions between components on coating performance:

[0011]

[0012] where f=[f1,f2,…,f n ] is the parameter vector of the coating formula, w i ,λ i ,β ij are model parameters;

[0013] S2. Secondly, the performance indicators are normalized and weighted, and the formula is used to smooth the weights and normalized performance indicators:

[0014]

[0015] where x=[x1,x2,…,x k ] is the performance indicator vector, v j is the weight of the jth performance indicator;

[0016] S3, then use the multi-objective genetic algorithm to optimize the formula:

[0017]

[0018] where p=[p1,p2,…,p m ] is the weighted and normalized performance index, u l is the optimization weight of the lth indicator, balancing multiple performance objectives by geometric mean;

[0019] S4. Finally, the production cost is integrated through the formula:

[0020]

[0021] Where c is the cost, p is the performance indicator vector, r l is the coefficient of the impact of performance on cost, and k is the parameter that adjusts the steepness of the curve.

[0022] Furthermore, the process of generating the initial population construction is as follows:

[0023] First, the ingredients and process parameters in the coating formula are selected, and the value range of each parameter is defined. The real number coding strategy is adopted, where each parameter f i Using real value direct encoding, the formula is expressed as:

[0024] C(f i )=a i +b i sin(πf i )

[0025] where a i and b i is the coefficient that adjusts the parameter range and distribution;

[0026] Next, we design the structure of the encoded genome to ensure that the value of each gene should fall within its defined parameter range, using the formula:

[0027]

[0028] Where F is the genome, f is the specific recipe parameter, and σ is the parameter that controls the distribution width;

[0029] Finally, a certain number of individuals are randomly generated as the initial population. Each individual consists of a set of encoded genes. This set of genes directly represents the complete coating formula and the parameters of its manufacturing process. The formula is:

[0030]

[0031] Where N is the population size, r i and k i are the parameters that regulate the influence of each gene.

[0032] Furthermore, the method for selecting an advantageous formula by the roulette wheel selection method comprises the following steps:

[0033] Firstly, the roulette wheel selection method is used to calculate the fitness score of each individual in the population and normalize it. The ratio of its fitness score to the total score is used as the selection probability to select the dominant formula.

[0034] Then, through the single-point or multi-point crossover method, a single or multiple breakpoints are randomly selected, and the paragraphs between these points of the two parent formulas are exchanged to generate a sub-formula with new characteristics;

[0035] Finally, the single or multiple ingredient ratios or process parameters in the newly generated formula are randomly mutated and modified according to probability and amplitude to explore new formulas.

[0036] Furthermore, the process of selecting the advantageous formula includes:

[0037] First, the individual's fitness score is calculated by a nonlinear function:

[0038]

[0039] where x k is the kth performance indicator, w k and α k It is the parameter that adjusts the impact of performance indicators;

[0040] Then the fitness score is normalized to a higher order using the exponential smoothing normalization formula:

[0041]

[0042] Where T is the parameter controlling the temperature, which is used to adjust the sensitivity of normalization;

[0043] Finally, the function is selected in integral form:

[0044]

[0045] where μ and σ are the mean and standard deviation parameters of the integral, respectively, which are used to adjust the shape of the selection probability distribution.

[0046] Furthermore, the step of generating a sub-formula with new characteristics comprises:

[0047] First, the single-point crossover operation is defined as randomly selecting a single breakpoint on the ingredient ratio encoding of the two parent recipes and using the formula:

[0048] X new =X1[:k]+X2[k:]

[0049] and

[0050] Y new =Y1[:k]+Y2[k:]

[0051] To generate two new sub-recipes, where k is a randomly selected breakpoint, and X1, X2 and Y1, Y2 are the codes of the parent recipes respectively;

[0052] Then the multi-point crossover operation is performed by selecting two or more breakpoints k1, k2, ..., k n And split the recipe code into multiple paragraphs according to the breakpoints, using the formula:

[0053]

[0054] where k0 = 0 and k n+1 = L, where L is the total length of the encoding, and paragraphs are exchanged from two parent recipes in an alternating manner to generate new child recipes.

[0055] Furthermore, the process of exploring new formula construction includes:

[0056] S1. First, according to the preset mutation probability p, use the adaptive mutation probability function:

[0057]

[0058] Determine whether each parameter v is mutated, where κ and θ are parameters for adjusting probability sensitivity and offset respectively;

[0059] S2. Then, for the continuous parameter f, the mixed normal perturbation formula is used:

[0060]

[0061] Among them, η controls the total amplitude of variation, ρ represents the sensitivity of the parameter to the formulation performance, and α n ,μ n ,σ n are the weight, mean and variance parameters of the mixed Gaussian model, and N is the number of Gaussian distributions that make up the mixed model;

[0062] S3. For discrete parameter g, use the high-order random jump formula:

[0063]

[0064] where δ is the maximum allowed step length, γ adjusts the nonlinear strength of the random jump, and R is a random number uniformly distributed in the interval [-1, 1].

[0065] Furthermore, the optimization control method of the coating performance is:

[0066] The first algorithm module developed uses a time- and parameter-based partial differential equation model to simulate the dynamic changes of nanomaterials during the reaction process:

[0067]

[0068] in represents the distribution state of nanomaterials, and represent the diffusion coefficient and reaction rate, respectively;

[0069] Then the genetic algorithm uses the fitness function:

[0070]

[0071] Optimize, where Represent the field distribution of temperature, pressure and chemical dosage respectively, ω T ,ω P ,ω C is the corresponding weight factor;

[0072] Finally, the dynamic control system uses the control function:

[0073]

[0074] Where s represents the real-time data obtained from the process monitoring system, k i and β i It is a parameter to adjust the reaction speed and accuracy.

[0075] Beneficial effects of the present invention:

[0076] 1. Enhanced formulation optimization capabilities: Through genetic algorithms, a large number of possible formulation combinations can be automatically explored to quickly find the optimal or near-optimal coating formulation. This method is more efficient than the traditional trial and error method and can significantly reduce development time and costs.

[0077] 2. Improve product performance: The algorithm ensures that the developed coating formula has excellent performance in actual application by comprehensively evaluating the key performance indicators of the coating, such as dispersion, viscosity, hardness and corrosion resistance. In addition, through the dynamic control system, the algorithm can adjust the production process according to real-time data to further optimize product performance.

[0078] 3. Increase production flexibility and adaptability: The introduction of genetic algorithms enables the coating production process to automatically adjust according to different production conditions and raw material batches, improving the flexibility and adaptability of the production process. This is especially important for responding to fluctuations in raw material supply and changes in market demand.

[0079] 4. Reduce production costs: By optimizing the formula and production parameters, the present invention can effectively reduce the waste of raw materials and optimize the use of energy and time, thereby reducing production costs. In addition, by reducing the number of trial and error in the production process, the cost of the research and development stage can also be significantly reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] Figure 1 The present invention is a flow chart of the production control method of nano-modified anti-corrosion coating based on genetic algorithm.

[0081] Figure 2 A flow chart of a method for selecting an advantageous formula using the roulette wheel selection method of the present invention. DETAILED DESCRIPTION

[0082] The specific implementation modes of the present invention will be described in detail below in conjunction with the accompanying drawings.

[0083] The production control method of nano-modified anti-corrosion coatings based on genetic algorithms first defines a multi-objective optimization fitness function to comprehensively evaluate the key performance indicators of the coating formula, including dispersion, viscosity, hardness and corrosion resistance, while also taking into account the impact of production costs; the purpose of the fitness function is to quantify the overall performance of each formula so that the genetic algorithm can find the optimal balance between multiple performance indicators of the coating. For example, the fitness function takes the form of a weighted sum, in which each performance indicator is assigned a different weight according to its importance in product performance. In addition, cost considerations are included as a negative impact factor to ensure that costs are reduced as much as possible without sacrificing performance. This method makes coating production not only pursue technical excellence, but also find the best economic balance between cost and performance.

[0084] Subsequently, real number coding is used to encode the ingredient ratios and process parameters in the coating formula, aiming to convert the various ingredients of the coating formula and parameters in the production process such as temperature, pressure, stirring speed, etc. into numerical forms that can be processed by the algorithm. The main advantage of real number coding is that it can directly represent the actual value of the variable, thereby allowing the genetic algorithm to operate in a continuous search space, which is more suitable for processing parameters that require high-precision adjustment than traditional binary coding. Next, the initial population is generated randomly or based on some heuristic method. The initial population consists of multiple different formulas, each of which represents a solution. The quality and diversity of the initial population have an important impact on the performance of the algorithm and the optimal solution that can ultimately be found. Therefore, when generating the initial population, it is necessary to ensure wide coverage to provide a good starting point for the iterative process of the genetic algorithm.

[0085] Then, the dominant formula is selected through the roulette selection method in genetic operations. This selection method is a commonly used natural selection simulation method, in which the selection probability of each formula is proportional to its fitness, ensuring that formulas with high fitness have a higher chance of being selected to enter the next generation, thereby increasing the probability of the algorithm converging to the optimal solution; in addition, in order to improve formula diversity and avoid premature convergence, the algorithm uses single-point or multi-point crossover methods, in which one or more crossover points are selected from two parent formulas, and then the segments between these points are exchanged to generate new formulas with parent characteristics; finally, in order to further explore the solution space and increase the genetic diversity of the population, the selected formula ingredient ratio or process parameters are randomly mutated, which includes making small-scale random adjustments to certain ingredient ratios or process parameters in the formula. Such mutations can introduce new genetic mutations and help find potentially effective formulas that have not been explored before. The entire process gradually optimizes the formula through continuous selection, crossover and mutation operations until the best coating formula is found.

[0086] In the final stage, an algorithm module is developed to control the size, shape and distribution of nanomaterials. This module uses the optimization results obtained from the genetic algorithm to guide the reaction conditions in the production process, such as temperature, pressure, and the amount of chemical reagents added, to ensure that these conditions can best affect the properties of nanomaterials. The genetic algorithm finds the formula and process parameters that lead to the optimal coating performance by simulating the process of natural selection and genetic variation. The algorithm module dynamically adjusts the manufacturing process based on these data to accurately control the growth and aggregation of nanoparticles, thereby achieving the purpose of controlling their final size, shape and uniform distribution. This control is crucial because these physical properties of nanomaterials are directly related to the corrosion resistance, wear resistance and adhesion properties of the coating. In this way, it can be ensured that each batch of coating production can meet the predetermined high standards, and the continuous optimization of coating performance and the consistency of product quality can be achieved.

[0087] Embodiment 1:

[0088] In this embodiment, a multi-objective optimization fitness function is constructed to evaluate the effects of each component in the formula and the interactions between the components on the coating properties through a high-dimensional nonlinear regression model.

[0089] It is assumed that a new nano-modified anti-corrosion coating is to be developed, and its formula includes three main components: resin (f1), curing agent (f2), and nano-additive (f3). The ratio of these three components directly affects the hardness, viscosity and corrosion resistance of the coating.

[0090] The recipes are evaluated using the following fitness function model:

[0091]

[0092] in:

[0093] α is a baseline performance parameter set to 0.5 and used to adjust the base coating properties.

[0094] w i ,λ i , and β ij is a model parameter and needs to be adjusted according to experimental data.

[0095] Set the parameters as follows:

[0096] w1=0.3, w2=0.2, w3=0.4: reflect the basic contribution of each component to the coating performance.

[0097] λ1=0.05, λ2=0.03, λ3=0.07: control the attenuation rate of the effect of component amount on performance.

[0098] β 12 =0.1,β 13 =0.05, β 23 =0.08: describes the influence of interaction between components.

[0099] If the parameter vector of the recipe is f = [0.1, 0.2, 0.05]:

[0100] P(f)=0.5+(0.3·exp(-0.05·0.1 2 )+0.2·exp(-0.03·0.2 2 )+0.4·exp(-0.07·0.05 2 ))

[0101] +(0.1 0.1 0.2+0.05 0.1 0.05+0.08 0.2 0.05)

[0102] P(f)=0.5+(0.3·exp(-0.0005)+0.2·exp(-0.0012)+0.4·exp(-0.000175))

[0103] +(0.02+0.00025+0.0008)

[0104] P(f)=0.5+(0.3·0.9995+0.2·0.9988+0.4·0.999825)+0.02105

[0105] P(f)=0.5+0.29985+0.19976+0.39993+0.02105

[0106] P(f)=1.4206

[0107] This calculation result shows that the coating performance score is 1.4206 under the given formulation parameters, indicating that the performance of this formulation is better than the baseline level. Using this method, the formulation parameters can be adjusted and the genetic algorithm can be used for iteration to find the optimal formulation that can maximize P(f).

[0108] Secondly, the performance indicators are normalized and weighted, and mathematical methods are used to smooth the weights and normalized performance indicators to ensure that each indicator is appropriately evaluated according to its contribution to the final product performance.

[0109] The formula will be used:

[0110]

[0111] where x=[x1,x2,…,x k ] is the performance indicator vector, v j is the weight of the jth performance indicator.

[0112] The performance indicators of the coating are set to include hardness x1, viscosity x2, and corrosion resistance x3. According to the experimental data, the actual measured values ​​of these performance indicators are as follows:

[0113] x1 = 0.8 (hardness, scale is 0 to 1)

[0114] x2 = 0.6 (viscosity, scale 0 to 1)

[0115] x3 = 0.9 (corrosion resistance, scale 0 to 1)

[0116] The weights are set according to the application scenarios and market demand of the coatings as follows:

[0117] v1=0.3 (weight of hardness)

[0118] v2 = 0.4 (weight of viscosity)

[0119] v3 = 0.3 (weight of corrosion resistance)

[0120] According to the set formula and parameters, the performance indicators are weighted and smoothed. This process includes the integration of the function of each performance indicator between 0 and 1, calculated as follows:

[0121]

[0122] By numerical integration, the calculated S(x) will provide a smoothed, normalized performance score that combines the contributions of each performance indicator weighted by its importance. For example, using numerical integration, calculations can be performed using appropriate mathematical software or programming environments such as MATLAB and Python to obtain a numerical score that represents the overall performance of the coating.

[0123] The next step involves using a multi-objective genetic algorithm to optimize the performance of the coating. In this stage, an optimization algorithm is used to balance multiple performance objectives.

[0124] The optimization formula is:

[0125]

[0126] where p=[p1,p2,…,p m ] is the performance index after normalization and weighting, u l is the optimization weight of the lth index. This formula balances multiple performance objectives by means of geometric mean in order to find a coating formula that performs well on all the performance indicators considered.

[0127] There are three performance indicators of interest: hardness p1, viscosity p2, and corrosion resistance p3. After normalization and weight assignment in the previous steps, the following weighted and normalized values ​​are obtained:

[0128] p1=0.85(hardness)

[0129] p2=0.75(viscosity)

[0130] p3=0.90(corrosion resistance)

[0131] The weights are set as:

[0132] u1=0.3(weight of hardness)

[0133] u2=0.4(weight of viscosity)

[0134] u3=0.3 (weight of corrosion resistance)

[0135] Substitute the optimization formula for calculation:

[0136]

[0137] P(p)=0.85 0.3 ×0.75 0.4 ×0.90 0.3

[0138] O(p)=0.9225×0.8707×0.9556

[0139] O(p)=0.769

[0140] This calculation result shows that after comprehensively considering the optimized weights of each performance index, the overall performance score is 0.769, indicating that this coating formulation performs well in meeting the requirements of hardness, viscosity and corrosion resistance.

[0141] Finally, cost considerations are integrated to ensure that economic benefits are taken into account while optimizing coating performance. The cost integration formula used reflects the complex relationship between production cost and performance, and adjusts the degree of influence of cost on the final decision through a parameterized method.

[0142] The cost integration formula is:

[0143]

[0144] Where c is the cost, p is the performance indicator vector, r l is the coefficient of the impact of performance on cost, and k is the parameter that adjusts the steepness of the curve. This formula simulates the trade-off between cost and performance in the form of a logistic function.

[0145] The performance indicators of the coating are set to include hardness p1, viscosity p2, and corrosion resistance p3, while the production costs considered include raw materials, labor and energy consumption. The following parameters are set:

[0146] p1=0.8(hardness)

[0147] p2=0.75(viscosity)

[0148] p3=0.9(corrosion resistance)

[0149] r1=0.2, r2=0.3, r3=0.5: reflect the influence coefficient of each performance index on cost.

[0150] c = 50 (unit: ten thousand yuan), which represents the basic cost of paint production.

[0151] k=0.1: used to adjust the steepness of the curve. A smaller k value means that the curve changes smoothly and is less sensitive to cost changes.

[0152] Substitute into the formula to calculate:

[0153]

[0154]

[0155] C(50,[0.8,0.75,0.9])≈0.9927

[0156] This calculation result shows that after comprehensively considering performance and cost, the obtained coating cost-effectiveness ratio is 0.9927, which is close to 1, indicating that the balance between cost and performance has reached a very close to optimal state.

[0157] Embodiment 2:

[0158] In this embodiment, generating an initial population construction provides the algorithm with a diverse starting point to explore the solution space. This process involves selecting appropriate coating composition and process parameters, and adopting a real number encoding strategy to ensure that each parameter can be flexibly adjusted to suit the needs of the genetic algorithm.

[0159] Generate the initial population construction process:

[0160] Step 1: Select ingredients and process parameters: First, the key ingredients in the coating formulation need to be determined, such as resin type, curing agent, stabilizer, pigment, filler and nano additives. In addition, process parameters such as stirring speed, curing temperature and time are also taken into consideration. The selection of these parameters will be based on the intended use and performance requirements of the coating.

[0161] Step 2: Define the range of values ​​for each parameter: Define a practical range of values ​​for each component and parameter. For example, the proportion of resin is between 0% and 50%, the proportion of curing agent is between 0% and 10%, and the stirring speed ranges from 100 to 1000 rpm. These ranges ensure the practical applicability and manufacturing feasibility of the parameters.

[0162] Step 3: Real number encoding strategy:

[0163] Each parameter f i Using real number coding, the coding formula is:

[0164] C(f i )=a i +b i sin(πf i )

[0165] Here, a i and b i is a coefficient that adjusts the range and distribution of the parameter. For example, if f i is the resin ratio, set a i =25 and b i = 25 to ensure that the encoding value is in the range of 0% to 50%. This encoding method not only simplifies the adjustment of parameters, but also increases the nonlinear characteristics in the encoding process, which helps to explore a wider solution space.

[0166] The resin ratio f1 needs to be coded, and the range of the resin ratio is set to 0% to 50%. The selected coefficients are a1=25 and b1=25, and the coding formula is:

[0167] C(f1)=25+25sin(πf1)

[0168] If the actual resin ratio f1 = 0.5 (where 0.5 represents a standardized value of 50%), then:

[0169] C(f1)=25+25sin(π×0.5)

[0170] C(f1)=25+25×1

[0171] C(f1)=50

[0172] This indicates that the encoding method can correctly reflect the values ​​of the parameters within their defined range. The method provides a systematic framework to adjust and optimize the paint formulation, and also ensures that the genetic algorithm can run effectively to find the most optimized paint formulation through real number encoding.

[0173] The structure of the encoded genome is then designed to ensure that the value of each gene should fall within its defined parameters. By using formulas to ensure this, not only is the accuracy of gene expression enhanced, but the search and development process of coating formulas is also optimized. This step focuses on ensuring the accuracy and practicality of gene expression, using mathematical models to simulate the variation in gene expression and its distribution within the allowed range.

[0174] The formula used when designing the genome structure is:

[0175]

[0176] Where F is the genome, f is the specific recipe parameter, and σ is the parameter that controls the distribution width, which is used to ensure that the gene variation is within a reasonable and practical range.

[0177] The setting is considering a specific formulation parameter - the proportion of curing agent, and its minimum value f min is 0.01(1%), the maximum value f max =0.1 (10%). Set the parameter σ that controls the distribution width to 0.02. It is necessary to calculate the expression distribution of the genome F at a certain value (such as 0.05, i.e., 5% curing agent ratio).

[0178] The practical application of the calculation formula can be completed by numerical integration method, the specific calculation is as follows:

[0179]

[0180] This integral can be solved using numerical integration methods, such as Simpson's rule or a software package with numerical integration capabilities. This calculation provides a probability density function for the gene within a defined range, which helps to evaluate how the variation in the curing agent ratio in the actual manufacturing process is controlled by the genetic algorithm optimization process.

[0181] Finally, the initial population is randomly generated to ensure that each individual generated can accurately reflect a complete set of coating formulations and parameters of its manufacturing process.

[0182] An initial population is randomly generated, and each individual consists of a set of encoded genes that directly represent the complete coating formula and its manufacturing parameters. The coating formula considered includes ingredients such as resin content, curing agent ratio, nanoparticle content, and parameters such as temperature and time during production.

[0183] Set the following parameters:

[0184] N=100: means generating a population of 100 individuals.

[0185] r i and k i are adjustment parameters, where r i The range of k is 0.1 to 0.9, i The range is 0.5 to 1.5 and is used to control the influence of the gene and the reaction rate.

[0186] Use the following formula to generate the recipe value for each individual:

[0187]

[0188] This formula reflects the influence of each gene in a nonlinear way by combining logarithmic and logical functions, ensuring the complexity and diversity of genetic variation.

[0189] Set the gene f of one of the individuals i The actual value of is 0.7 (set to represent the percentage of resin content), select r i = 0.5 and k i =1.0 to calculate:

[0190]

[0191] P(n)=0.5log(1+0.66822)

[0192] P(n)=0.5log(1.66822)

[0193] P(n)=0.5×0.5108

[0194] P(n)=0.2554

[0195] The use of this method not only enhances the precision of gene expression, but also optimizes the search and development process of coating formulas, ensuring the scientific nature of the method and the feasibility of practical application.

[0196] Embodiment 3:

[0197] In this embodiment, a roulette wheel selection method is used to select the dominant formula to ensure that the formula with high fitness has a greater probability of being selected to enter the next generation, thereby optimizing the performance of the entire coating formula.

[0198] The roulette wheel selection method selects based on the relative fitness of individuals. Individuals with higher fitness scores will have a higher probability of being selected. Considering multiple performance indicators of the coating, a nonlinear function is used to calculate the fitness score of each individual.

[0199] The performance indicators of the coating are set to include hardness x1, viscosity x2, corrosion resistance x3, and cost efficiency x4. Assign weights and adjustment parameters to each performance indicator:

[0200] w1=0.25,w2=0.25,w3=0.25,w4=0.25—The weights of each performance indicator are equal, indicating that all performance indicators are equally important to the overall performance.

[0201] α1=1.0,α2=1.5,α3=0.5,α4=0.8—parameters that control the period of the sine function and the behavior of the logarithmic function, and adjust the influence of each performance indicator.

[0202] The fitness score is calculated as:

[0203]

[0204] The specific performance data are set as follows:

[0205] x1=0.8(hardness)

[0206] x2=0.6(viscosity)

[0207] x3=0.9(corrosion resistance)

[0208] x4=0.4(cost efficiency)

[0209] Substitute into the formula to calculate:

[0210]

[0211] This calculation can be done using math software or a programming language. Let the results be 0.56, 0.45, 0.62, and 0.38. Sum these values ​​to get the total of F(i).

[0212] In this way, the fitness score of each individual is calculated and used in the roulette wheel selection method, ensuring that better formulations have a higher chance of entering the next generation, which helps to continuously optimize the coating formulation.

[0213] Next, high-order normalization of the fitness score is performed to accurately evaluate the relative advantages of each recipe and provide a standardized comparison basis for the selection process. Using the exponential smoothing normalization formula, the sensitivity of the normalization is adjusted by controlling the parameter T, which helps to balance the competitive pressure in the population and avoid the premature solidification of the local optimal solution. The implementation of this formula can ensure that the algorithm continues to evolve effectively over multiple generations, rather than relying solely on individual individuals with high fitness.

[0214] The normalization formula used is:

[0215]

[0216] Where T is a control parameter, often referred to as "temperature", which is used to adjust the sensitivity of the normalization process. A lower T value will make individuals with high fitness scores stand out significantly, while a higher T value will reduce the score difference and increase the diversity of the population.

[0217] Assuming the population N is 100 individuals, the following are some assumed settings of fitness scores F(i) and T and their calculation examples:

[0218] The fitness scores F(i) of the four random individuals are set to 1.2, 0.8, 2.1, and 1.5 respectively.

[0219] Set the control temperature T=0.5.

[0220] To calculate the normalized fitness scores of these individuals, first calculate the index score of each individual:

[0221] exp(F(1) / T)=exp(1.2 / 0.5)=exp(2.4)≈11.023

[0222] exp(F(2) / T)=exp(0.8 / 0.5)=exp(1.6)≈4.953

[0223] exp(F(3) / T)=exp(2.1 / 0.5)=exp(4.2)≈66.686

[0224] exp(F(4) / T)=exp(1.5 / 0.5)=exp(3.0)≈20.086

[0225] Next, calculate the denominator of the normalized score:

[0226]

[0227] The normalized score for each individual is:

[0228]

[0229] Through this calculation example, we can see how to normalize the score of each individual according to its fitness score and the control parameter T. This normalization method not only fairly reflects the performance of each individual relative to other individuals in the population, but also allows the algorithm operator to control the competitive pressure and diversity within the population by adjusting the value of T, thereby optimizing the evolutionary effect of the entire population.

[0230] Finally, the selection function is used, especially in the optimization process to select the individuals with the most potential for reproduction. The selection function in the form of an integral can accurately control the probability of individuals being selected, ensuring that individuals with high fitness have a greater chance of participating in the next generation of formula innovation.

[0231] The selection function is defined as:

[0232]

[0233] Where μ and σ are the mean and standard deviation parameters of the integral, respectively, which are used to adjust the shape of the selection probability distribution. This method uses the properties of the normal distribution to adjust the selection probability of each individual, making the process more detailed and precise.

[0234] Assume that the individuals in the population have been evaluated and normalized, with the following steps and parameter settings:

[0235] The normalized fitness score S(i) of the individual is set to 0.85.

[0236] We set μ = 0.5 and σ = 0.1 to make the distribution center at 0.5 and the standard deviation of 0.1, indicating that the distribution is relatively concentrated, highlighting the selection probability of individuals with medium and high fitness.

[0237] Calculate the integral:

[0238]

[0239] This integral can be solved by numerical integration methods, such as using the integral function in mathematical software or programming languages ​​(such as Python's SciPy library). Through this calculation, the probability of an individual being selected can be obtained, which is based on the individual's normalized fitness score and the specified normal distribution parameters.

[0240] Embodiment 4:

[0241] In this embodiment, the crossover operation is used to generate a sub-formula with new characteristics. Through the single-point or multi-point crossover method, the advantages of two parent formulas can be effectively combined to explore new characteristics, thereby enhancing the diversity and innovation of the entire population.

[0242] 1. Randomly select breakpoints: Randomly select a single breakpoint on the ingredient ratio encoding of the parent recipe. This breakpoint k will determine how to take a portion from each of the two parent recipes to create the child recipe.

[0243] 2. Generate a new sub-recipe: Use the formula:

[0244] X new =X1[:k]+X2[k:]

[0245] and

[0246] Y new =Y1[:k]+Y2[k:]

[0247] Here, X1, X2 and Y1, Y2 are the codes of two sets of parent recipes respectively. This method ensures that the parts taken from each parent recipe can be combined into a complete new recipe.

[0248] The code for setting up two parent recipes is as follows:

[0249] Parent formula 1X1: 0.2, 0.3, 0.5, 0.7, 0.9 (representing different ingredient ratios)

[0250] Parent formula 2X2: 0.1, 0.4, 0.6, 0.8, 0.2

[0251] Set the randomly selected breakpoint k = 3. According to the above formula, the new sub-formulas X and Y will consist of the following parts:

[0252] new new

[0253] X new =X1[:3]+X2[3:]=[0.2,0.3,0.5]+[0.8,0.2]=[0.2,0.3,0.5,0.8,0.2]

[0254] Y new =Y1[:3]+Y2[3:] (If Y1 and Y2 have actual values, the same operation will be performed)

[0255] Through the single-point crossover operation, the new child recipe inherits the characteristics of both parent recipes while also exhibiting new performance characteristics. The implementation of this approach not only improves the exploration ability of the genetic algorithm, but also increases the innovation of the recipe. The randomness of the selected breakpoints ensures the diversity of the operation, while the practice of evenly taking parts from each parent recipe ensures the maintenance of genetic diversity.

[0256] Then, a multi-point crossover operation is performed to exchange multiple paragraphs from two parent recipes by selecting multiple breakpoints, thereby generating sub-recipes with new features. This method can improve the innovation of recipes and enhance the efficiency of the algorithm when exploring complex recipe spaces.

[0257] 1. Select breakpoints: Randomly select two or more breakpoints k1, k2, ..., k on the ingredient ratio codes of the two parent formulas. n . These breakpoints split the parent recipe code into multiple paragraphs.

[0258] 2. Generate new sub-recipes: Use the formula

[0259]

[0260] where k0 = 0 and k n+1 = L (L is the total length of the code). This formula creates a new sub-recipe by alternately taking paragraphs from two parent recipes at breakpoint locations.

[0261] The code for setting up two parent recipes is as follows:

[0262] Parent formula 1X: 0.2, 0.3, 0.5, 0.7, 0.9, 0.4, 0.2

[0263] Parent formula 2Y: 0.1, 0.4, 0.6, 0.8, 0.2, 0.5, 0.3

[0264] The randomly selected breakpoints are set to k1=2 and k2=5 (the total length of the recipe code is set to L=7).

[0265] According to the above breakpoints and formula, the new sub-formula Z is calculated as follows:

[0266] new

[0267] k0=0,k3=7

[0268] Z new =(-1) 1 (X[0:2]+Y[2:5])+(-1) 2 (X[2:5]+Y[5:7])+(-1) 3 (X[5:7]+Y[7:7])

[0269] Z new =(-1)([0.2,0.3]+[0.6,0.8,0.2])+(-1) 2 ([0.5,0.7,0.9]+[0.5,0.3])

[0270] +(-1) 3 ([0.4,0.2]+[])

[0271] Z new =[-0.8,-1.1,-0.2]+[1.0,1.0,1.2]+[-0.4,-0.2]

[0272] Z new =[0.2,-0.1,1.0,1.0,1.0,0.1,0.0]

[0273] Multi-point crossover operations achieved complex recipe mixing through breakpoint and alternating selection strategies, ensuring the increase of genetic diversity and the introduction of new recipe characteristics.

[0274] Embodiment 5:

[0275] In this embodiment, by randomly mutating the proportions of one or more components or process parameters in the newly generated formula, the diversity and innovation of the coating formula can be achieved. The adaptive mutation probability function used in this process allows the algorithm to flexibly adjust the mutation probability according to the current state of the population, ensuring that the mutation operation can maintain the diversity of the population without causing excessive mutation and deviating from the optimization direction.

[0276] 1. Determine the mutation probability: First, based on the preset mutation probability p of each parameter v, use the adaptive mutation probability function to decide whether to mutate the parameter. This probability is not fixed, but dynamically adjusted according to the importance of the parameter and the current population status.

[0277] 2. Adaptive mutation probability function:

[0278]

[0279] Among them, κ and θ are parameters for adjusting probability sensitivity and offset. These two parameters allow the sensitivity and offset of the variation to be adjusted to meet the optimization needs of different stages.

[0280] Set the following parameters:

[0281] Preset mutation probability p = 0.05 (5% probability of mutation)

[0282] κ=10:Higher value increases the response sensitivity of the function

[0283] θ = 0.05: ensures that when p approaches the preset threshold, the mutation probability quickly approaches 50%

[0284] Consider a specific process parameter v (e.g. stirring speed). It is necessary to calculate whether this parameter should be varied in the current iteration. The calculation is performed using the above parameters and formula:

[0285]

[0286] This means that the parameter has a 50% probability of mutating, which is an example value. By adjusting κ and θ, this probability can be adjusted according to actual conditions.

[0287] In this way, the mutation probability of each parameter can be flexibly adjusted to ensure that the population can effectively explore the unknown solution space in the process of pursuing the optimal solution while maintaining appropriate stability. This adaptive mutation probability function is highly customizable and can be adjusted according to the optimization needs of different stages.

[0288] Then the continuous parameters are varied in an optimized formula. The mixed normal perturbation formula can accurately control the variation process, making the variation more detailed and goal-oriented. This method allows different variation modes to coexist by introducing a mixed Gaussian model, thus adapting to complex optimization needs.

[0289] 1. Define parameters: For continuous parameters f, such as the proportion of a specific chemical component in the paint, the mixed normal perturbation formula is used for variation.

[0290] 2. Mixed normal disturbance formula:

[0291]

[0292] in:

[0293] η controls the total amplitude of variation, for example 0.1 (indicating that the maximum amplitude of variation is 10% of the original parameter).

[0294] ρ represents the sensitivity of the parameter to the formulation performance, and its value range is usually between 0 and 1, such as 0.5.

[0295] α n is the weight of each normal distribution in the mixture, ensuring that it sums to 1.

[0296] μ n and σ n are the mean and standard deviation of each normal distribution. The mean can vary around 0, and the standard deviation is usually small, such as 0.05.

[0297] N is the number of Gaussian distributions that make up the mixture model. For example, 3 means that three normal distributions with different parameters are used to construct the variation.

[0298] The ratio of the paint components is set to f=0.25, hoping to explore new formula space through mutation.

[0299] Setting parameters: η = 0.1, ρ = 0.5, N = 3

[0300] Distribution parameters: α1=0.5, μ1=0, σ1=0.02; α2=0.3, μ2=0.01, σ2=0.01; α3=0.2, μ3=-0.01, σ3=0.01

[0301] calculate:

[0302] f′=0.25+0.1·(1-0.5 2 )·(0.5·N(0,0.02 2 )+0.3·N(0.01,0.01 2 )+0.2·N(-0.01,0.01 2 ))

[0303] Set the randomly generated normal distribution value to N(0,0.02 2 )=0.005,N(0.01,0.01 2 )=0.012,

[0304] N(-0.01,0.01 2 )=-0.008:

[0305] f ′ =0.25+0.1·0.75·(0.5·0.005+0.3·0.012+0.2·-0.008)

[0306] f ′ =0.25+0.1·0.75·(0.0025+0.0036-0.0016)

[0307] f ′ =0.25+0.1·0.75·0.0045

[0308] f ′ =0.25+0.0003375

[0309] f ′ =0.2503375

[0310] Through the above calculations, we can see how the mixed normal perturbation formula finely adjusts the ratio of ingredients in the paint formula. This strategy allows the algorithm to gradually approach the optimal solution through careful mutation operations while maintaining sufficient randomness to avoid falling into the local optimum. n ,μ n , and σ n Parameters such as , R & D team can flexibly adjust the variation strategy according to specific formulation requirements and performance sensitivity.

[0311] Finally, for the mutation operation involving discrete parameters, a high-order random jump formula is used. It is suitable for parameters with discrete value ranges and fixed specific value intervals, such as the proportion or frequency of use of certain specific additives in the paint formula. In this way, new values ​​of parameters can be effectively explored, the diversity of the population can be increased, and a better formula can be found.

[0312] 1. Define parameters: For discrete parameters g in the coating formula, such as the number of times an additive is used or the type number of a specific chemical component.

[0313] 2. High-order random jump formula:

[0314]

[0315] in:

[0316] δ is the maximum allowed step length, for example, if the parameter is the number of times the additive is used, the maximum step length is 3.

[0317] γ adjusts the nonlinear strength of random jumps, and smaller values ​​enhance the nonlinear effect, such as 2.

[0318] R is a random number uniformly distributed in the interval [-1,1].

[0319] The discrete parameter g is set to 10 (this is set to represent the number of times a certain additive is used).

[0320] Set parameters: δ=3, γ=2.

[0321] Set the randomly generated number R = -0.64:

[0322]

[0323] g ′ =10+(-1)·floor(0.8·3)

[0324] g ′ =10-floor(2.4)

[0325] g ′ =10-2

[0326] g ′ =8

[0327] Through the above calculations, we can see the specific operation and effect of using the high-order random jump formula to mutate discrete parameters. This method introduces nonlinear factors and randomness, allowing the algorithm to maintain exploratory nature while avoiding excessive focus on local optimal solutions.

[0328] Embodiment 6:

[0329] In this embodiment, the optimization control of coating performance uses a partial differential equation model to simulate the dynamic changes of nanomaterials during the reaction process through the algorithm module. This model can provide deep insights into how to accurately control the distribution of materials during the manufacturing process, thereby optimizing the performance of the final product.

[0330] Equation definition:

[0331]

[0332] This partial differential equation describes how nanomaterials diffuse and react as a function of time and other parameters during the reaction.

[0333] Variable explanation:

[0334] represents the distribution state of the nanomaterial, where t represents time, Represents other parameters that affect the behavior of nanomaterials, such as temperature, pressure, etc.

[0335] is the diffusion coefficient, which determines the rate and mode of diffusion of nanomaterials and is usually related to the physical properties of the material and external conditions.

[0336] is the reaction rate, which affects how quickly and how strongly a material reacts.

[0337] Suppose that a new type of anti-corrosion coating is being developed and the diffusion and reaction behavior of a specific nano-additive needs to be simulated. To achieve this, the following parameters need to be set:

[0338] set up This indicates that the diffusion rate is low, which is due to the high viscosity of the coating matrix.

[0339] set up This indicates that the reaction rate is moderate.

[0340] Consider a one-dimensional example where u(t,x) depends only on position x and time t, and consider an initial value problem, e.g., at t=0, the nanoadditive is concentrated only at x=0:

[0341] u(0,x)=δ(x)

[0342] Solving this equation using numerical methods (such as the finite difference method) can simulate how the nano-additive diffuses and reacts in the coating over time. The above partial differential equation can be solved iteratively by setting the time step and spatial grid.

[0343] The implementation of this method ensures that the genetic algorithm-based nano-modified anti-corrosion coating production control method is not only advanced in theory, but also can effectively guide the production process in practical applications. By adjusting parameters such as D and λ, it can flexibly respond to different manufacturing conditions and material properties and optimize the overall performance of the coating.

[0344] The design of the fitness function is the core of the algorithm optimization process, which determines how to evaluate and select the optimal formula. By integrating the field distribution of key process parameters such as temperature, pressure and chemical dosage, the fitness function can comprehensively consider the impact of multiple factors on coating performance.

[0345] The formula of fitness function is:

[0346]

[0347] in:

[0348] Represent the field distribution of temperature, pressure and chemical dosage respectively, these parameters reflect the environmental and material conditions at various stages in the coating production process.

[0349] ω T ,ω P ,ω C are the corresponding weight factors, which indicate the importance of different parameters on the coating performance.

[0350] Ω represents the spatial domain of the entire paint production environment or container, and and Represent the position in space and a specific parameter point respectively.

[0351] Consider a coating production process where temperature, pressure, and chemical dosage have different distributions in a specific spatial region. Set the following parameters:

[0352] ω T =0.5,ω P =0.3,ω c =0.2—indicates that temperature is the most important factor for coating performance, followed by pressure and chemical dosage.

[0353] The set temperature T in the production area changes from 20℃ to 100℃, the pressure P changes from 1atm to 10atm, and the chemical dosage C changes from 0.1kg / m 3 Up to 1.0kg / m 3 .

[0354] We can assume that Ω is the unit cube, and each point in the paint reaction container All have different values ​​of T, P, and C. If the integral is calculated over the entire spatial domain, numerical methods are required, such as finite element analysis or Monte Carlo simulation.

[0355] By adjusting the weight factor ω T ,ω P , and ω C, the optimization strategy can be flexibly adjusted according to different production requirements and material properties, enabling the R&D team to more effectively explore and utilize the diversity of the formulation space, thereby accelerating the development cycle of new formulations.

[0356] Finally, a dynamic control system is introduced to ensure quality control and performance optimization during the production process. By using control functions, the system can dynamically adjust the production process according to the changes in various parameters monitored in real time, so as to achieve the predetermined product quality and performance indicators.

[0357] Definition of control function:

[0358]

[0359] in:

[0360] s represents real-time data obtained from the process monitoring system, which can include real-time readings of key parameters such as temperature, pressure, chemical dosage, etc.

[0361] k i and β i It is the adjustment coefficient, which is used to control the reaction speed and accuracy.

[0362] k i Affects the strength of the control reaction, with larger values ​​meaning a stronger response to real-time changes.

[0363] β i Controls the damping term to stabilize the system and prevent excessive oscillations.

[0364] The key parameters that need to be controlled during the coating production process include temperature and pressure. Set the following parameters:

[0365] n=2 (control temperature and pressure parameters)

[0366] k1=0.1,k2=0.05—Control intensity of temperature and pressure

[0367] β1=0.01,β2=0.02—damping coefficients of temperature and pressure

[0368] The temperature measured by the real-time monitoring system at a certain moment is set to s1 = 100°C (the temperature deviates from the target temperature by 20 degrees), and the pressure is set to s2 = 5atm (the pressure deviates from the target pressure by 1atm). and

[0369] According to the control function:

[0370]

[0371] By adjusting k i and βi Parameters such as temperature and humidity can flexibly respond to different production conditions and material properties, optimize the overall performance of the coating, and thus achieve higher market competitiveness and product quality.

[0372] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A method for controlling the production of nano-modified anti-corrosion coatings based on genetic algorithms, characterized in that The following steps are involved: First, the fitness function of multi-objective optimization is defined to comprehensively evaluate the dispersion, viscosity, hardness and corrosion resistance of the coating formula, and consider the production cost; Then, the component ratios and process parameters in the coating formula are encoded using real number coding, and the initial population is generated; Then, the superior formula is selected through the roulette wheel selection method in genetic operation, the formula diversity is improved by using single-point or multi-point crossover method, and the ingredient ratio or process parameters are randomly mutated to obtain a new formula; Finally, an algorithm module was developed to control the size, shape and distribution of nanomaterials, and the reaction conditions were adjusted according to the optimization results of the genetic algorithm to achieve the optimization of coating performance; The method for selecting an advantageous formula by the roulette wheel selection method comprises the following steps: Firstly, the roulette wheel selection method is used to calculate the fitness score of each individual in the population and normalize it. The ratio of its fitness score to the total score is used as the selection probability to select the dominant formula. Then, through the single-point or multi-point crossover method, a single or multiple breakpoints are randomly selected, and the paragraphs between these points of the two parent formulas are exchanged to generate a sub-formula with new characteristics; Finally, the single or multiple ingredient ratios or process parameters in the newly generated formula are randomly mutated and modified according to probability and magnitude; The process of selecting the advantageous formula includes: First, the individual's fitness score is calculated by a nonlinear function: where x k is the kth performance indicator, w k and α k It is the parameter that adjusts the impact of performance indicators; Then the fitness score is normalized to a higher order using the exponential smoothing normalization formula: Where T is the parameter controlling the temperature, which is used to adjust the sensitivity of normalization; Finally, the function is selected in integral form: Where μ and σ are the mean and standard deviation parameters of the integral, respectively, which are used to adjust the shape of the selection probability distribution; The steps of generating a sub-formula with new characteristics include: First, the single-point crossover operation is defined as randomly selecting a single breakpoint on the ingredient ratio encoding of the two parent recipes and using the formula: X new =X1[:k]+X2[k:] and Y new =Y1[:k]+Y2[k:] To generate two new sub-recipes, where k is a randomly selected breakpoint, and X1, X2 and Y1, Y2 are the codes of the parent recipes respectively; Then the multi-point crossover operation is performed by selecting two or more breakpoints k1, k2, ..., k n And split the recipe code into multiple paragraphs according to the breakpoints, using the formula: where k0 = 0 and k n+1 = L, L is the total length of the code, swapping paragraphs from two parent recipes in an alternating manner to generate new child recipes; The process of exploring new formula construction includes: S1. First, according to the preset mutation probability p, use the adaptive mutation probability function: Determine whether each parameter v is mutated, where κ and θ are parameters for adjusting probability sensitivity and offset respectively; S2. Then, for the continuous parameter f, the mixed normal perturbation formula is used: Among them, η controls the total amplitude of variation, ρ represents the sensitivity of the parameter to the formulation performance, and α n ,μ n ,σ n are the weight, mean and variance parameters of the mixed Gaussian model, and N is the number of Gaussian distributions that make up the mixed model; S3. For discrete parameter g, use the high-order random jump formula: where δ is the maximum allowed step length, γ adjusts the nonlinear strength of the random jump, and R is a random number uniformly distributed in the interval [-1, 1].

2. The method for controlling the production of nano-modified anti-corrosion coatings based on genetic algorithm according to claim 1, characterized in that The fitness function construction of the multi-objective optimization includes: S1. First, a high-dimensional nonlinear regression model is constructed to capture the effects of individual components and interactions between components on coating performance: where f=[f1,f2,…,f n ] is the parameter vector of the coating formula, w i ,λ i ,β ij is a model parameter; α is a baseline performance parameter used to adjust the base coating performance; S2. Secondly, the performance indicators are normalized and weighted, and the formula is used to smooth the weights and normalized performance indicators: where x=[x1,x2,…,x k ] is the performance indicator vector, v j is the weight of the jth performance indicator; S3, then use the multi-objective genetic algorithm to optimize the formula: where p=[p1,p2,…,p m ] is the weighted and normalized performance index, u l is the optimization weight of the lth indicator, balancing multiple performance objectives by geometric mean; S4. Finally, the integrated production cost is calculated by the formula: Where c is the cost, p is the performance indicator vector, r l is the coefficient of the impact of performance on cost, and k is the parameter that adjusts the steepness of the curve.

3. The method for controlling the production of nano-modified anti-corrosion coatings based on genetic algorithm according to claim 1, characterized in that The process of generating the initial population construction is as follows: First, the ingredients and process parameters in the coating formula are selected, and the value range of each parameter is defined. The real number coding strategy is adopted, where each parameter f i Using real value direct encoding, the formula is expressed as: C(f i )=a i +b i sin(πf i ) where a i and b i is the coefficient that adjusts the parameter range and distribution; Next, we design the structure of the encoded genome to ensure that the value of each gene should fall within its defined parameter range, using the formula: Where F is the genome, f is the specific recipe parameter, and σ is the parameter that controls the distribution width; Finally, a certain number of individuals are randomly generated as the initial population. Each individual consists of a set of encoded genes. This set of genes directly represents the complete coating formula and the parameters of its manufacturing process. The formula is: Where N is the population size, r i and k i are the parameters that regulate the influence of each gene.

4. The method for controlling the production of nano-modified anti-corrosion coatings based on genetic algorithm according to claim 1, characterized in that The optimization control method of the coating performance: The first algorithm module developed uses a time- and parameter-based partial differential equation model to simulate the dynamic changes of nanomaterials during the reaction process: in represents the distribution state of nanomaterials, and represent the diffusion coefficient and reaction rate, respectively; Then the genetic algorithm uses the fitness function: Optimize, where Represent the field distribution of temperature, pressure and chemical dosage respectively, ω T ,ω P ,ω C is the corresponding weight factor; and and Respectively represent the position and specific parameter point in space; Finally, the dynamic control system uses the control function: Where s represents the real-time data obtained from the process monitoring system, k i and β i It is a parameter to adjust the reaction speed and accuracy.

Images