Industrial coating formula intelligent optimization method and system
By extracting and analyzing the feature vectors related to coating raw materials, scenario requirements, and performance, and using the gradient descent algorithm to optimize industrial coating formulations, the problems of long cycle and high cost in traditional methods are solved, achieving efficient and economical formulation optimization.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- FOSHAN HUANG GUAN CHEM IND CO LTD
- Filing Date
- 2026-01-30
- Publication Date
- 2026-05-01
AI Technical Summary
Traditional industrial coating formulation design relies on trial and error and expert experience, resulting in long experimental cycles, high costs, and knowledge gaps, making it impossible to optimize for user needs.
By extracting the raw material feature vector, scenario requirement feature vector, and performance-related feature vector of industrial coatings, the gradient descent algorithm is used for iterative optimization to output the optimal formula parameters. Combined with performance penalty terms and cost minimization objectives, intelligent formula optimization is achieved.
It shortens the industrial coating formulation development cycle, reduces cost waste, improves formulation development efficiency, and meets specific performance requirements.
Smart Images
Figure CN121963936A_ABST
Abstract
Description
A method and system for intelligent optimization of industrial coating formulations Technical Field
[0001] This invention relates to the field of coating technology, and more specifically, to a method and system for intelligent optimization of industrial coating formulations. Background Technology
[0002] Industrial coatings are chemical coating materials with film-forming substances, pigments, solvents, and additives as their basic components. They are used in various fields, including automotive, marine, and corrosion protection, and mainly include water-based coatings and powder coatings. Traditionally, industrial coating formulations are often based on fixed formulas, resulting in relatively fixed performance and making it impossible to optimize and adjust the formulation to meet specific user needs.
[0003] Current industrial coating formulation design mainly relies on trial and error and expert experience, which has significant drawbacks, including: 1. Long experimental cycle, with a single formulation verification requiring process steps such as mixing, coating, curing, and performance testing; 2. Strong reliance on experience: senior formulation engineers are scarce, and there is a risk of knowledge gaps in knowledge transmission.
[0004] Therefore, it is necessary to develop an intelligent optimization method for industrial coating formulations to improve the efficiency of formulation development. Summary of the Invention
[0005] Based on this, in order to improve the efficiency of industrial coating formulation development, this invention provides an intelligent optimization method and system for industrial coating formulations, the specific technical solution of which is as follows:
[0006] A method for intelligent optimization of industrial coating formulations includes the following steps: acquiring basic raw material parameters, target application scenario requirement parameters, and historical formulation performance data of industrial coatings; preprocessing the collected basic raw material parameters, scenario requirement parameters, and historical formulation performance data to extract raw material feature vectors, scenario requirement feature vectors, and performance-related feature vectors; and iteratively optimizing the raw material ratio based on the raw material feature vectors, scenario requirement feature vectors, and performance-related feature vectors, with the goal of minimizing raw material costs and the constraint of meeting scenario performance requirements, to output the optimal formulation parameters.
[0007] The intelligent optimization method for industrial coating formulations extracts raw material feature vectors, scenario requirement feature vectors, and performance-related feature vectors. Based on these vectors, it iteratively optimizes the raw material ratio with the goal of minimizing raw material costs and the constraint of meeting scenario performance requirements. This method outputs optimal formulation parameters, enabling cost minimization while meeting performance requirements in industrial coating formulation optimization. This shortens the development cycle, reduces cost waste, and improves the efficiency of industrial coating formulation development.
[0008] Preferably, the basic raw material parameters include resin type, filler particle size, and additive ratio; the target application scenario requirements parameters include corrosion resistance, temperature resistance, and adhesion requirements; and the historical formulation performance data includes test data of past mass production formulations.
[0009] Preferably, the specific method for iteratively optimizing the raw material ratio includes the following steps: obtaining the total cost of the raw materials in the formula, the performance index requirement threshold, and the predicted value of the performance index, and obtaining a performance penalty term based on the performance index requirement threshold and the predicted value of the performance index; obtaining the core objective function based on the total cost of the raw materials in the formula and the performance penalty term, constructing a formula optimization model based on the core objective function, and iteratively optimizing the raw material ratio based on the gradient descent algorithm.
[0010] Preferably, the core objective function is expressed as follows: ;in, The vectors represent the raw material proportions, the total cost of the raw materials in the formula, and the performance penalty term, respectively, where c represents the raw material unit price vector. The k-th term is represented by the predicted value of the performance metric, the threshold value of the performance metric requirement, and the performance constraint penalty weight, respectively. This is a penalty function used to impose a non-linear penalty for underperformance, where k represents the number of performance metrics.
[0011] Preferably, the intelligent optimization method for industrial coating formulations further includes the following steps: converting the optimal formulation parameters into standardized formulation instructions and associating them with the corresponding performance prediction report.
[0012] An intelligent optimization system for industrial coating formulations, used to implement the aforementioned intelligent optimization method for industrial coating formulations, includes: a data acquisition module for acquiring basic raw material parameters, target application scenario requirement parameters, and historical formulation performance data of industrial coatings; a feature extraction module for preprocessing the acquired basic raw material parameters, scenario requirement parameters, and historical formulation performance data to extract raw material feature vectors, scenario requirement feature vectors, and performance-related feature vectors; and a formulation optimization module for iteratively optimizing the raw material ratio based on the raw material feature vectors, scenario requirement feature vectors, and performance-related feature vectors, with the goal of minimizing raw material costs and the constraint of meeting scenario performance requirements, and outputting optimal formulation parameters; wherein, the modules achieve bidirectional data interaction through a data bus, the output of the feature extraction module is connected to the input of the formulation optimization module, and the output of the formulation optimization module is connected to the input of the result output module.
[0013] Preferably, the basic raw material parameters include resin type, filler particle size, and additive ratio; the target application scenario requirements parameters include corrosion resistance, temperature resistance, and adhesion requirements; and the historical formulation performance data includes test data of past mass production formulations.
[0014] Preferably, the formula optimization module includes: a performance penalty acquisition unit, used to acquire the total cost of formula raw materials, the performance index requirement threshold, and the predicted value of the performance index, and to acquire a performance penalty term based on the performance index requirement threshold and the predicted value of the performance index; and an optimization model construction unit, used to acquire a core objective function based on the total cost of formula raw materials and the performance penalty term, and to construct a formula optimization model based on the core objective function; wherein, the formula optimization model is based on the gradient descent algorithm to iteratively optimize the raw material ratio.
[0015] Preferably, the optimization model building unit is based on the formula Construct the core objective function; where, The vectors represent the raw material proportions, the total cost of the raw materials in the formula, and the performance penalty term, respectively, where c represents the raw material unit price vector. The k-th term is represented by the predicted value of the performance metric, the threshold value of the performance metric requirement, and the performance constraint penalty weight, respectively. This is a penalty function used to impose a non-linear penalty for underperformance, where k represents the number of performance metrics.
[0016] Preferably, the intelligent optimization system for industrial coating formulations further includes a result output module, used to convert the optimal formulation parameters into standardized formulation instructions and associate them with the corresponding performance prediction report. Attached Figure Description
[0017] The invention will be further understood from the following description taken in conjunction with the accompanying drawings. The components in the drawings are not necessarily drawn to scale, but rather the emphasis is on illustrating the principles of the embodiments. In different views, the same reference numerals designate corresponding parts.
[0018] Figure 1 is a schematic diagram of the overall process of an intelligent optimization method for industrial coating formulations in an embodiment of the present invention; Figure 2 is a schematic diagram of the specific method for iterative optimization of raw material ratios in an embodiment of the present invention; Figure 3 is a schematic diagram of the overall structure of an intelligent optimization system for industrial coating formulations in an embodiment of the present invention. Detailed Implementation
[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to its embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not limit the scope of protection of the invention.
[0020] It should be noted that when an element is referred to as being "fixed to" another element, it can be directly attached to the other element or there may be an intervening element. When an element is referred to as being "connected to" another element, it can be directly connected to the other element or there may be an intervening element. The terms "vertical," "horizontal," "left," "right," and similar expressions used herein are for illustrative purposes only and do not represent the only possible implementation.
[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.
[0022] In this invention, "first" and "second" do not represent a specific quantity or order, but are merely used to distinguish names.
[0023] As shown in Figure 1, an embodiment of the present invention provides an intelligent optimization method for industrial coating formulations, including the following steps: S1, obtaining basic raw material parameters of industrial coatings, target application scenario requirement parameters, and historical formulation performance data.
[0024] Specifically, the basic raw material parameters include, but are not limited to, resin type, filler particle size, and additive ratio; the target application scenario requirements parameters include, but are not limited to, corrosion resistance, temperature resistance, and adhesion requirements; and the historical formula performance data includes, but is not limited to, the test data of past mass production formulas.
[0025] S2 preprocesses the collected basic raw material parameters, scenario requirement parameters, and historical formula performance data to extract raw material feature vectors, scenario requirement feature vectors, and performance correlation feature vectors.
[0026] S3, based on the raw material feature vector, scenario requirement feature vector, and performance correlation feature vector, aims to minimize raw material costs and meets scenario performance requirements, iteratively optimizes the raw material ratio, and outputs the optimal formula parameters.
[0027] Specifically, the raw material proportioning vector is defined as follows: , Let n be the percentage of the i-th raw material and n be the number of raw material types. Then, the cost minimization objective is expressed as: ;in Let the unit price of raw materials be a vector. Let be the unit price of the i-th raw material.
[0028] Performance requirement constraints (mm performance indicators) are expressed as follows: ,k=1,2,…,m. These represent the predicted value of the performance index of the kth item and the required threshold value of the performance index of the kth item, respectively, such as corrosion resistance ≥ level 5.
[0029] As a preferred technical solution, as shown in Figure 2, the specific method for iteratively optimizing the raw material ratio includes the following steps: S31, obtaining the total cost of the raw materials in the formula, the performance index requirement threshold, and the predicted value of the performance index, and obtaining the performance penalty term based on the performance index requirement threshold and the predicted value of the performance index. S32, obtaining the core objective function based on the total cost of the raw materials in the formula and the performance penalty term, constructing a formula optimization model based on the core objective function, and iteratively optimizing the raw material ratio based on the gradient descent algorithm.
[0030] For example, the core objective function is expressed as ;in, The vectors represent the raw material proportions, the total cost of the raw materials in the formula, and the performance penalty term, respectively, where c represents the raw material unit price vector. The k-th term is represented by the predicted value of the performance metric, the threshold value of the performance metric requirement, and the performance constraint penalty weight, respectively. This is a penalty function used to impose a non-linear penalty for underperformance, where k represents the number of performance metrics.
[0031] Specifically, the total cost of the formulation raw materials is equal to the sum of the products of the proportion of each type of raw material and its corresponding unit price, used to quantify the economics of the formulation, with the goal of minimizing it. Performance index predictions can be obtained based on a neural network model, which takes raw material feature vectors and raw material proportion vectors as inputs and outputs predicted performance index values.
[0032] Performance requirement thresholds are determined based on scenario-specific feature vectors, such as a performance requirement threshold of 1200h for marine coatings. Penalty function. Defined as , Indicates the performance difference. denoted as the curvature adjustment factor, typically set to 0.1. When z>0, the penalty term increases cubically with z, forcing the optimization away from catastrophic recipes and infeasible regions through superlinear growth; while the quadratic term dominates with minor non-compliance, and the cubic term penalizes severe non-compliance.
[0033] The performance constraint penalty weight of the k-th term is dynamically updated using an adaptive adjustment mechanism, with the update rule being: . Let represent the performance constraint penalty weights of the k-th term in iteration t and t+1, respectively.
[0034] Specifically, performance not meeting standards When the performance constraint penalty weight of the k-th term is increased exponentially to strengthen the penalty, the performance constraint penalty weight of the k-th term remains unchanged when the performance meets the target to avoid over-optimization. This is the sensitivity coefficient, which can be set according to the application scenario of the industrial coating. For example, it can be set to β=0.8 for marine coatings and β=0.3 for toy coatings.
[0035] The core objective function includes performance requirements. This hard constraint, through a performance penalty term incorporated into the objective function, transforms constrained optimization into an unconstrained problem. In the initial stage, if the formulation is far from the feasible region (performance is severely substandard), the penalty term dominates the optimization direction, prioritizing performance improvement. In the later stage, once performance meets the standards, the total cost of the formulation's raw materials takes the lead, finely optimizing economic efficiency to dynamically balance cost and performance.
[0036] The performance constraint penalty weights are dynamically updated using an adaptive adjustment mechanism. The cubic penalty term α will generate a steep gradient in the severely unmet area, helping to escape local optima and avoid local optima.
[0037] After iteratively optimizing the raw material ratio based on the gradient descent algorithm to obtain the optimal formula parameters, the optimal formula parameters can be converted into standardized formula instructions and associated with the corresponding performance prediction report.
[0038] In summary, the intelligent optimization method for industrial coating formulations extracts raw material feature vectors, scenario requirement feature vectors, and performance-related feature vectors. Based on these vectors, and with the goal of minimizing raw material costs and the constraint of meeting scenario performance requirements, the method iteratively optimizes the raw material ratios and outputs optimal formulation parameters. This method can minimize costs while meeting performance requirements in industrial coating formulation optimization, thereby shortening the development cycle, reducing cost waste, and improving the efficiency of industrial coating formulation development.
[0039] In one embodiment, it can also be based on a function. Improve the gradient descent direction. Specifically, These represent the raw material ratio at the t-th iteration and the updated raw material ratio vector, which is the ratio for the next iteration, respectively.
[0040] Let be the learning rate (step size) at the t-th iteration. It should be noted that in improved algorithms, the learning rate is usually adaptive and may be dynamically adjusted according to the iteration process (e.g., decreasing as the number of iterations increases) to ensure convergence. Cost function At point The gradient at that point. Total raw material cost function. It is usually a linear function, and its gradient is... It points in the direction where costs increase the fastest, so it needs to be updated in its negative direction to reduce costs. Represents the penalty function At point The gradient at a given point points in the direction that increases the penalty, so this gradient should be subtracted during the update to reduce the penalty.
[0041] By weighted summation of the gradients of all m performance constraint penalty terms, all performance constraints are considered simultaneously, and the impact of each constraint is adjusted according to its weight. Thus, cost can be optimized while satisfying performance constraints within the gradient descent framework, specifically including: 1. This optimization is driven by negative gradients. The direction of the distribution sector should be adjusted to reduce costs.
[0042] 2. Performance-constrained driving. This is achieved through negative penalty gradients. The algorithm adjusts the direction of the algorithm to meet all performance requirements. When a performance level falls below a threshold, the corresponding penalty function map is non-zero and its direction is the direction that improves that performance; therefore, updates will be made in the direction that improves that performance.
[0043] 3. Dynamic trade-off mechanism. The penalty weight of each performance constraint is adaptively adjusted during the iteration process. When a certain performance is severely unsatisfactory, its corresponding performance constraint penalty weight will increase, making the performance constraint have a greater impact on the update direction; when the performance is satisfied, the performance constraint penalty weight may decrease or remain unchanged, thus allowing cost optimization to dominate the update direction.
[0044] The second derivative information of the performance prediction model can be incorporated into gradient calculation to avoid local oscillations, i.e.: Furthermore, the convergence efficiency of high-dimensional non-convex spaces is improved by utilizing the Hessian matrix.
[0045] When competing performance metrics exist (such as high temperature resistance vs. low viscosity), a multi-objective optimization strategy is adopted to construct the objective vector. Pareto solution sets are generated using the improved NSGA-II framework.
[0046] As shown in Figure 3, an embodiment of the present invention provides an intelligent optimization system for industrial coating formulations, which is used to implement the intelligent optimization method for industrial coating formulations. The system includes a data acquisition module, a feature extraction module, and a formulation optimization module.
[0047] The data acquisition module is used to acquire basic raw material parameters of industrial coatings, target application scenario requirements parameters, and historical formula performance data. The feature extraction module is used to preprocess the acquired basic raw material parameters, scenario requirement parameters, and historical formula performance data to extract raw material feature vectors, scenario requirement feature vectors, and performance-related feature vectors. The formula optimization module is used to iteratively optimize the raw material ratio based on the raw material feature vectors, scenario requirement feature vectors, and performance-related feature vectors, with the goal of minimizing raw material costs and the constraint of meeting scenario performance requirements, and output the optimal formula parameters. The modules interact bidirectionally through a data bus. The output of the feature extraction module is connected to the input of the formula optimization module, and the output of the formula optimization module is connected to the input of the result output module.
[0048] Basic raw material parameters include resin type, filler particle size, and additive ratio; target application scenario requirements include corrosion resistance, temperature resistance, and adhesion requirements; historical formulation performance data includes test data of past mass production formulations.
[0049] Raw material feature vector Including resin type (One-hot), filler particle size (Gaussian distribution encoding), and additive ratio (normalized vector), etc., scenario requirement feature vector. Including corrosion resistance threshold Temperature resistance Adhesion Standardized values, performance-related feature vectors, etc. This includes the latent space mapping of historical formulation performance data (such as salt spray resistance time and adhesion rating). Joint features can be generated through feature concatenation and attention weighting to obtain... W is a learnable weight matrix, and the weights are dynamically assigned through a gated attention mechanism. This avoids a rigid combination of raw material and demand characteristics, and adaptively highlights key constraints, such as increasing the weight of resin type characteristics when the corrosion resistance threshold is extremely high.
[0050] Specifically, for raw material feature vectors, the basic raw material parameters can first be cleaned and standardized, such as removing outliers and normalizing the proportions; the resin type can be structured and encoded, and the filler particle size distribution can be characterized using a Geometric Mixture Model (GMM) encoding. Extraction of scenario-related feature vectors involves standardizing parameters such as corrosion resistance, temperature resistance, and adhesion, and automatically weighting them according to the application scenario to generate a requirement vector. Extraction of performance-related feature vectors involves: cross-source data alignment based on patented formulation performance data and mass production formulation test reports, followed by knowledge graph construction and embedding, and finally, deep encoding of performance data using an adversarial autoencoder (AAE) structure to generate corresponding vectors.
[0051] As a preferred technical solution, the formula optimization module includes a performance penalty acquisition unit and an optimization model construction unit.
[0052] The performance penalty acquisition unit is used to acquire the total cost of raw materials in the formula, the performance index requirement threshold, and the predicted value of the performance index, and to acquire the performance penalty term based on the performance index requirement threshold and the predicted value of the performance index; the optimization model construction unit is used to acquire the core objective function based on the total cost of raw materials in the formula and the performance penalty term, and to construct the formula optimization model based on the core objective function; wherein, the formula optimization model is based on the gradient descent algorithm to iteratively optimize the raw material ratio.
[0053] For example, the optimization model building unit is based on the formula Construct the core objective function; where, The vectors represent the raw material proportions, the total cost of the raw materials in the formula, and the performance penalty term, respectively, where c represents the raw material unit price vector. The k-th term is represented by the predicted value of the performance metric, the threshold value of the performance metric requirement, and the performance constraint penalty weight, respectively. This is a penalty function used to impose a non-linear penalty for underperformance, where k represents the number of performance metrics.
[0054] The predicted value of the performance index for the k-th term is defined as a higher-order nonlinear mapping of the eigenvectors, i.e. .in, represents the tensor product of the raw material proportion vector and the raw material feature vector, characterizing the component interaction effect. ⊙ represents the element-wise multiplication of the scenario demand feature vector and the performance-related feature vector, to strengthen the constraint of historical experience. It is a three-layer neural network with LeakyReLU activation function.
[0055] Specifically, the total cost of the formulation raw materials is equal to the sum of the products of the proportion of each type of raw material and its corresponding unit price, used to quantify the economics of the formulation, with the goal of minimizing it. Performance index predictions can be obtained based on a neural network model, which takes raw material feature vectors and raw material proportion vectors as inputs and outputs predicted performance index values.
[0056] Performance requirement thresholds are determined based on scenario-specific feature vectors, such as a performance requirement threshold of 1200h for marine coatings. Penalty function. Defined as , Indicates the performance difference. denoted as the curvature adjustment factor, typically set to 0.1. When z>0, the penalty term increases cubically with z, forcing the optimization away from catastrophic recipes and infeasible regions through superlinear growth; while the quadratic term dominates with minor non-compliance, and the cubic term penalizes severe non-compliance.
[0057] The performance constraint penalty weight of the k-th term is dynamically updated using an adaptive adjustment mechanism, with the update rule being: . Let represent the performance constraint penalty weights of the k-th term in iteration t and t+1, respectively.
[0058] Specifically, performance not meeting standards When the performance constraint penalty weight of the k-th term is increased exponentially to strengthen the penalty, the performance constraint penalty weight of the k-th term remains unchanged when the performance meets the target to avoid over-optimization. This is the sensitivity coefficient, which can be set according to the application scenario of the industrial coating. For example, it can be set to β=0.8 for marine coatings and β=0.3 for toy coatings.
[0059] The core objective function includes performance requirements. This hard constraint, through a performance penalty term incorporated into the objective function, transforms constrained optimization into an unconstrained problem. In the initial stage, if the formulation is far from the feasible region (performance is severely substandard), the penalty term dominates the optimization direction, prioritizing performance improvement. In the later stage, once performance meets the standards, the total cost of the formulation's raw materials takes the lead, finely optimizing economic efficiency to dynamically balance cost and performance.
[0060] The performance constraint penalty weights are dynamically updated using an adaptive adjustment mechanism. The cubic penalty term α will generate a steep gradient in the severely unmet area, helping to escape local optima and avoid local optima.
[0061] The aforementioned intelligent optimization system for industrial coating formulations also includes a results output module. This module converts the optimal formulation parameters into standardized formulation instructions and associates them with corresponding performance prediction reports.
[0062] Specifically, based on a preset mapping function between formulation parameters and formulation instructions, the system uses the optimal formulation parameters as input to the mapping function and outputs standardized formulation instructions. The format of these standardized formulation instructions includes, but is not limited to, JSON / OPCUA and XML. The performance prediction report includes, but is not limited to, physical properties (adhesion, abrasion resistance), chemical properties (salt spray resistance, solvent resistance), and economics (total cost of formulation raw materials).
[0063] This invention aims to minimize raw material costs and uses performance requirements of various scenarios as constraints for iterative optimization. By integrating cost-driven, performance-compensated, and physical-constrained factors, it intelligently optimizes industrial coating formulations based on the gradient descent algorithm, which helps to shorten the formulation optimization cycle and reduce costs.
[0064] In summary, the intelligent optimization system for industrial coating formulations extracts raw material feature vectors, scenario requirement feature vectors, and performance-related feature vectors. Based on these vectors, and with the goal of minimizing raw material costs and the constraint of meeting scenario performance requirements, it iteratively optimizes the raw material ratios and outputs optimal formulation parameters. This system can minimize costs while meeting performance requirements in industrial coating formulation optimization, thereby shortening the development cycle, reducing cost waste, and improving the efficiency of industrial coating formulation development.
[0065] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0066] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.
Claims
1. A method for intelligent optimization of industrial coating formulations, characterized in that, The process includes the following steps: acquiring basic raw material parameters for industrial coatings, target application scenario requirements, and historical formulation performance data; preprocessing the collected basic raw material parameters, scenario requirements, and historical formulation performance data to extract raw material feature vectors, scenario requirement feature vectors, and performance-related feature vectors; and iteratively optimizing the raw material ratio based on the raw material feature vectors, scenario requirement feature vectors, and performance-related feature vectors, with the goal of minimizing raw material costs and the constraint of meeting scenario performance requirements, to output the optimal formulation parameters.
2. The intelligent optimization method for industrial coating formulations as described in claim 1, characterized in that, Basic raw material parameters include resin type, filler particle size, and additive ratio; target application scenario requirements include corrosion resistance, temperature resistance, and adhesion requirements; historical formulation performance data includes test data of past mass production formulations.
3. The intelligent optimization method for industrial coating formulations as described in claim 2, characterized in that, The specific method for iteratively optimizing the raw material ratio includes the following steps: obtaining the total cost of the raw materials in the formula, the performance index requirement threshold, and the predicted value of the performance index, and obtaining a performance penalty term based on the performance index requirement threshold and the predicted value of the performance index; obtaining the core objective function based on the total cost of the raw materials in the formula and the performance penalty term, constructing a formula optimization model based on the core objective function, and iteratively optimizing the raw material ratio based on the gradient descent algorithm.
4. The intelligent optimization method for industrial coating formulations as described in claim 3, characterized in that, The core objective function is expressed as follows: ;in, The vectors represent the raw material proportions, the total cost of the raw materials in the formula, and the performance penalty term, respectively, where c represents the raw material unit price vector. The k-th term is represented by the predicted value of the performance metric, the threshold value of the performance metric requirement, and the performance constraint penalty weight, respectively. This is a penalty function used to impose a non-linear penalty for underperformance, where k represents the number of performance metrics.
5. The intelligent optimization method for industrial coating formulations as described in claim 4, characterized in that, It also includes the following steps: converting the optimal formulation parameters into standardized formulation instructions and associating them with the corresponding performance prediction report.
6. An intelligent optimization system for industrial coating formulations, used to implement the intelligent optimization method for industrial coating formulations as described in any one of claims 1-5, characterized in that, The intelligent optimization system for industrial coating formulations includes: a data acquisition module for acquiring basic raw material parameters, target application scenario requirement parameters, and historical formulation performance data; a feature extraction module for preprocessing the acquired basic raw material parameters, scenario requirement parameters, and historical formulation performance data to extract raw material feature vectors, scenario requirement feature vectors, and performance-related feature vectors; and a formulation optimization module for iteratively optimizing the raw material ratio based on the raw material feature vectors, scenario requirement feature vectors, and performance-related feature vectors, with the goal of minimizing raw material costs and the constraint of meeting scenario performance requirements, and outputting the optimal formulation parameters. The modules interact bidirectionally via a data bus, with the output of the feature extraction module connected to the input of the formulation optimization module, and the output of the formulation optimization module connected to the input of the result output module.
7. The intelligent optimization system for industrial coating formulations as described in claim 6, characterized in that, Basic raw material parameters include resin type, filler particle size, and additive ratio; target application scenario requirements include corrosion resistance, temperature resistance, and adhesion requirements; historical formulation performance data includes test data of past mass production formulations.
8. The intelligent optimization system for industrial coating formulations as described in claim 7, characterized in that, The formulation optimization module includes: a performance penalty acquisition unit, used to acquire the total cost of formulation raw materials, performance indicator requirement thresholds, and predicted performance indicator values, and to acquire performance penalty terms based on the performance indicator requirement thresholds and predicted performance indicator values; and an optimization model construction unit, used to acquire the core objective function based on the total cost of formulation raw materials and performance penalty terms, and to construct a formulation optimization model based on the core objective function; wherein, the formulation optimization model is based on the gradient descent algorithm to iteratively optimize the raw material ratio.
9. The intelligent optimization system for industrial coating formulations as described in claim 8, characterized in that, Optimize model building units according to formula Construct the core objective function; where, The vectors represent the raw material proportions, the total cost of the raw materials in the formula, and the performance penalty term, respectively, where c represents the raw material unit price vector. The k-th term is represented by the predicted value of the performance metric, the threshold value of the performance metric requirement, and the performance constraint penalty weight, respectively. This is a penalty function used to impose a non-linear penalty for underperformance, where k represents the number of performance metrics.
10. The intelligent optimization system for industrial coating formulations as described in claim 9, characterized in that, Also includes: The results output module is used to convert the optimal formula parameters into standardized formula instructions and associate them with the corresponding performance prediction report.