Aluminum-plastic composite tape process parameter intelligent recommendation method and system

Abstract

The application discloses an aluminum-plastic composite tape process parameter intelligent recommendation method and system, and relates to the technical field of aluminum-plastic composite tape intelligent manufacturing. The method comprises the following steps: obtaining physical property data of a material to be processed before preparation; judging whether the material is a new material based on a sample database, and calculating a physical property similarity coefficient if the material is a new material; dynamically constructing an adaptive quality prediction plug-in based on the principle of ensemble learning, combining a sample quality predictor library and the physical property similarity coefficient; and taking the maximization of product quality as a target, iteratively optimizing and searching in a heat-combining process parameter threshold space by using the plug-in, obtaining an adaptive process parameter space, and taking the adaptive process parameter space as a recommendation result. The application effectively solves the problem of poor prediction accuracy of new materials caused by the dependence of traditional methods on manual trial and error and insufficient historical data, realizes fast, accurate and automatic recommendation of process parameters, significantly improves production efficiency and product quality stability, and reduces development cost.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing technology for aluminum-plastic composite strips, specifically to an intelligent recommendation method and system for process parameters of aluminum-plastic composite strips. Background Technology

[0002] Aluminum-plastic composite tape is a key composite material widely used in communications, cables, aerospace and other fields. It is usually made of aluminum foil and plastic film through a thermal bonding process. The quality of the final product, such as peel strength, corrosion resistance and mechanical properties, directly determines the shielding effect and service life of the cable.

[0003] Currently, the determination of production process parameters for aluminum-plastic composite strips mainly relies on the experience of process engineers and repeated trial and error. When processing new materials, multiple small-batch trial productions, tests, and adjustments are required, resulting in long development cycles, high material and energy consumption, and high costs. Although some studies have attempted to use data modeling for process optimization, they generally rely on a large amount of similar historical data, leading to poor predictive effects and insufficient generalization ability for new materials lacking samples, making it difficult to achieve accurate recommendations for process parameters. Summary of the Invention

[0004] This invention addresses the technical problems of low efficiency and high cost caused by relying on manual trial and error in existing technologies, and the poor prediction effect of data-driven methods on process parameters of new materials due to the lack of historical data. It provides an intelligent recommendation method and system for process parameters of aluminum-plastic composite strips.

[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:

[0006] In a first aspect, the present invention provides a method for intelligent recommendation of process parameters for aluminum-plastic composite strips, comprising:

[0007] Before preparing the aluminum-plastic composite strip, obtain the material property data of the material to be processed;

[0008] Based on the sample material property database, determine whether the material to be processed is a new material according to the material property data. If it is a new material, obtain several property similarity coefficients.

[0009] Based on the principle of ensemble learning, an adaptive quality prediction plugin is constructed according to the sample quality predictor library and the aforementioned physical property similarity coefficients;

[0010] Based on the aforementioned quality prediction plugin, with the goal of maximizing the quality of the aluminum-plastic composite strip, and using the threshold of the thermal composite process parameters as the optimization space, the thermal composite process parameters are iteratively optimized and searched according to the material property data to obtain the adaptive thermal composite process parameter space.

[0011] The space of adaptive thermal composite process parameters is used as the recommended process parameters for the material to be processed.

[0012] Secondly, the present invention provides an intelligent recommendation system for process parameters of aluminum-plastic composite strips, comprising:

[0013] The material property data acquisition module is used to acquire the material property data of the material to be processed before the aluminum-plastic composite strip is prepared.

[0014] The new material judgment and similarity coefficient calculation module is used to determine whether the material to be processed is a new material based on the sample material property database and the material property data. If it is a new material, it obtains several property similarity coefficients.

[0015] The quality prediction plugin building module is used to build an adapted quality prediction plugin based on the principle of ensemble learning, according to the sample quality predictor library and the aforementioned physical property similarity coefficients.

[0016] The process parameter optimization search module is used to perform iterative optimization search of the thermal composite process parameters based on the material property data, with the goal of maximizing the quality of the aluminum-plastic composite strip and the thermal composite process parameter threshold as the optimization space, to obtain the adaptive thermal composite process parameter space.

[0017] The process parameter recommendation result output module is used to use the adapted thermal composite process parameter space as the process parameter recommendation result for the material to be processed.

[0018] The beneficial effects of this invention are:

[0019] Compared to existing technologies, this invention firstly effectively utilizes historical sample data through physical property similarity coefficients, overcoming the problems of poor adaptability to new materials and reliance on large amounts of similar data in traditional data-driven methods; secondly, it dynamically constructs an adaptive quality prediction plugin based on ensemble learning, improving the generalization ability and accuracy of the process parameter prediction model; thirdly, it replaces manual trial and error with intelligent iterative optimization search, reducing the number of trial productions, shortening the development cycle, and reducing material and energy consumption; finally, it achieves rapid, accurate, and automated recommendation of process parameters for aluminum-plastic composite strip production, effectively saving production costs while improving product quality stability. Attached Figure Description

[0020] Figure 1 A flowchart illustrating an intelligent recommendation method for process parameters of aluminum-plastic composite strip provided by the present invention;

[0021] Figure 2 This is a schematic diagram of the structure of an intelligent recommendation system for process parameters of aluminum-plastic composite strip provided by the present invention.

[0022] In the attached diagram, the components represented by each number are as follows:

[0023] Module 11 for acquiring material property data, Module 12 for judging new materials and calculating similarity coefficients, Module 13 for constructing quality prediction plugins, Module 14 for optimizing and searching process parameters, and Module 15 for outputting process parameter recommendation results. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0026] In the description of this invention, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed herein.

[0027] Example 1, as Figure 1 As shown, this embodiment of the invention provides a method for intelligent recommendation of process parameters for aluminum-plastic composite strips, including:

[0028] S10: Obtain material property data of the material to be processed before the aluminum-plastic composite tape is prepared;

[0029] Obtain the materials to be processed for preparing aluminum-plastic composite tape, and the material property data of the materials to be processed. The materials to be processed include aluminum foil, plastic film and adhesive. The material property data of aluminum foil includes at least thickness, density, specific heat capacity, thermal conductivity, surface tension, yield strength and elongation. The material property data of plastic film includes plastic film type and plastic film property information. The material property data of adhesive includes adhesive type and adhesive property information.

[0030] Specifically, in the production process of aluminum-plastic composite strips, the materials to be processed refer to all the basic raw materials that need to be prepared before entering the thermal bonding process, including aluminum foil, plastic film, and adhesives. Aluminum foil, as an extremely thin sheet of rolled aluminum, is the structural substrate and functional core of the aluminum-plastic composite strip. Its thickness, density, specific heat capacity, thermal conductivity, surface tension, yield strength, and elongation must be obtained. The material properties of the aluminum foil directly affect its heat transfer behavior, deformation characteristics, and interfacial bonding effect with the plastic film during the thermal bonding process.

[0031] Plastic film forms the polymer layer of the composite tape, typically serving as an outer protective layer or an inner contact layer for cable cores. Common materials include polyethylene (PE), polyethylene terephthalate (PET), polypropylene (PP), or cast polypropylene (CPP). Plastic films provide insulation, corrosion resistance, moisture resistance, and abrasion resistance; their type and properties must be clearly defined. Type is a qualitative classification, while properties are quantitative performance indicators such as thickness, density, melting point, glass transition temperature, and modulus. Both together determine the processing window of the plastic film and affect the flexibility and heat resistance of the final composite tape.

[0032] Furthermore, adhesives are substances used to firmly bond aluminum foil and plastic film through chemical and physical actions, providing durable and strong interlayer adhesion to ensure that the composite tape does not delaminate during subsequent processing and use. It is necessary to specify the type of adhesive, including polyurethane adhesives, acrylic adhesives, maleic anhydride grafted polymers, etc., as well as property information such as solid content, viscosity, pot life, curing temperature range, activation energy of the curing reaction, and surface tension. Adhesive property data are crucial for predicting whether it can achieve good adhesion under specific process conditions.

[0033] S20: Based on the sample material property database, determine whether the material to be processed is a new material according to the material property data. If it is a new material, obtain several property similarity coefficients.

[0034] Specifically, based on the sample material property database, it is determined whether the material to be processed is a new material according to the material property data. If it is a new material, several property similarity coefficients are obtained, including:

[0035] A sample material property database is constructed based on historical aluminum-plastic composite strip processing records of similar thermal composite equipment, wherein the sample material property database includes several sample material property data.

[0036] The material property data and the material property data of the several samples are compared and similarity is performed to obtain several property similarity coefficients.

[0037] If the number of coefficients with a value of 1 among the plurality of physical property similarity coefficients is not 0, then the material to be processed is an old material; if the number of coefficients with a value of 1 among the plurality of physical property similarity coefficients is 0, then the material to be processed is a new material.

[0038] First, a sample material property database is constructed based on historical aluminum-plastic composite strip processing records from similar thermal laminating equipment. Similar thermal laminating equipment refers to thermal laminating units with the same or similar heating methods, pressure application mechanisms, roller structures, and control precision. The historical aluminum-plastic composite strip processing records from similar thermal laminating equipment are derived from verified production process data reports, process monitoring logs, and corresponding quality inspection reports accumulated during the long-term operation of the equipment. Based on these historical aluminum-plastic composite strip processing records, a sample material property database is constructed. This database includes several sample material property data sets, which can systematically store and associate material parameters from different batches with their corresponding process settings, facilitating rapid retrieval, similarity comparison, and reuse of process knowledge.

[0039] Secondly, a similarity comparison was performed on the material property data and the material property data of several samples to obtain several property similarity coefficients. Specifically, the obtained material property data was compared with the material property data of several samples in the sample material property database. By calculating the comprehensive similarity between the material to be tested and each sample material in the multidimensional property feature space, several property similarity coefficients were obtained. The property similarity coefficient is a dimensionless index used to quantify the overall similarity of the physical properties between the material to be tested and a certain historical sample material. The coefficient ranges from [0,1], with values ​​closer to 1 indicating greater similarity in physical properties. The property similarity coefficient is used to evaluate the similarity between the new material and historically processed materials and provides a weighted basis for the subsequent construction of an adaptive quality prediction model.

[0040] Furthermore, if the number of coefficients with a value of 1 among the several physical property similarity coefficients is not zero, then the material to be processed is an old material; if the number of coefficients with a value of 1 among the several physical property similarity coefficients is zero, then the material to be processed is a new material. Specifically, a number of coefficients with a value of 1 among the several physical property similarity coefficients not being zero indicates that there are one or more historical material records in the current sample database that are completely identical to the physical properties of the material to be processed, thus determining that the material to be processed is an old material; a number of coefficients with a value of 1 among the several physical property similarity coefficients not being zero indicates that there are no historical samples in the current database that are completely identical to the physical properties of the material to be processed, thus determining that the material to be processed is a new material.

[0041] If the material is identified as an old material, it indicates that the material or a material with highly similar properties has been successfully processed in the past, and its corresponding optimal process parameters can be directly used or used after minor adjustments. If the material is identified as a new material, it indicates that the current combination of material properties exceeds the scope of existing historical experience, and existing process parameters cannot be directly used. A subsequent intelligent recommendation process needs to be initiated to determine the appropriate process.

[0042] S30: Based on the principle of ensemble learning, an adaptive quality prediction plugin is constructed according to the sample quality predictor library and the aforementioned physical property similarity coefficients;

[0043] Specifically, based on the principle of ensemble learning, an adaptive quality prediction plugin is constructed according to the sample quality predictor library and the aforementioned physical property similarity coefficients, including:

[0044] Based on deep learning, several sample quality predictors are constructed according to the material property data of the several samples, and a sample quality predictor library is obtained. The output of the quality predictor is peel strength, appearance quality coefficient and barrier performance parameter.

[0045] The ratio of the physical property similarity coefficient to the first constant is rounded down to the number of predictors to be selected. Several numbers of predictors to be selected are calculated based on the several physical property similarity coefficients, wherein the first constant is 0.05.

[0046] Based on the principles of ensemble learning and the average fusion mechanism, an adaptive quality prediction plugin is constructed according to the number of sample quality predictors and the number of predictors selected.

[0047] If a material is identified as a new material, and historical data for new materials is lacking, making it impossible to directly call existing process parameters or use a traditional single model for accurate quality prediction, then it is necessary to build an adapted quality prediction plugin based on the principle of ensemble learning, according to a sample quality predictor library and several physical property similarity coefficients.

[0048] First, based on deep learning, several sample quality predictors are constructed using material property data from several samples, resulting in a sample quality predictor library. Deep learning is a machine learning method that automatically learns deep features and complex mapping relationships in data through multi-layer nonlinear neural network structures. Models include deep neural networks (DNNs), convolutional neural networks (CNNs), and recurrent neural networks (RNNs), used to mine high-order nonlinear correlations between material properties, process parameters, and final quality indicators from large amounts of historical process data. Several sample quality predictors are constructed based on the material property data of several samples. Specifically, the sample material property data and different thermal composite process parameters are used as input features, and the actual measured peel strength, appearance quality coefficient, and barrier performance parameters corresponding to the sample under a specific process are used as training labels. Multiple deep neural network models are independently trained through supervised learning. The outputs of the quality predictors are peel strength, appearance quality coefficient, and barrier performance parameters.

[0049] For example, a deep neural network (DNN) is chosen to construct the sample quality predictor. A DNN is a deep feedforward neural network with a multilayer perceptron, consisting of an input layer, multiple hidden layers, and an output layer. It uses multiple nonlinear transformations to learn the complex mapping relationship between input features and output targets. Specifically, the physical properties of the sample material and the thermal composite process parameters are concatenated into a one-dimensional feature vector as the network input. This vector undergoes layer-by-layer feature transformation and abstraction through multiple hidden layers, and finally, the predicted values ​​of peel strength, appearance quality coefficient, and barrier performance parameters are mapped by the output layer. Backpropagation algorithms, such as gradient descent, are used to iteratively optimize the network weights to minimize the loss function (e.g., mean squared error) between the predicted values ​​and the true quality labels. Training stops when the model's performance on the validation set converges, ultimately resulting in a high-precision sample quality predictor.

[0050] The trained predictors are stored and compiled into a unified library, ultimately resulting in a sample quality predictor library.

[0051] Furthermore, the ratio of the physical property similarity coefficient to the first constant is rounded down to determine the number of predictors selected. Several predictor selection numbers are calculated based on several physical property similarity coefficients, where the first constant is 0.05. The first constant refers to the step size threshold for weight allocation, determining the unit scale required to convert the change in similarity into an integer increment of weight. A first constant of 0.05 means that for every 0.05 increase in similarity, the weight of the corresponding predictor in the ensemble model, i.e., the number of selected replicas, increases by approximately one unit. Higher similarity leads to higher predictive reliability of the corresponding model; therefore, a larger number of predictors needs to be allocated to enhance their influence in the ensemble decision, thereby improving the overall predictive reliability. The first constant controls the granularity of weight allocation; a smaller first constant results in finer granularity and greater sensitivity to weight changes. The ratio of the physical property similarity coefficient to the first constant is a continuous proportional value, representing the theoretical relative weight strength of the historical predictor. Rounding down the ratio of the physical property similarity coefficient to the first constant to determine the number of predictors selected represents discretizing the theoretical weights into the specific number of votes or replicas assigned to the predictor during actual ensemble integration. The number of predictors selected is calculated based on several physical property similarity coefficients, thereby determining the specific weight or contribution of each historical sample quality predictor in the subsequent integration and fusion.

[0052] Furthermore, based on the principles of ensemble learning and the averaging fusion mechanism, an adapted quality prediction plugin is constructed according to a number of sample quality predictors and a selection of predictor numbers. Specifically, the ensemble learning principle refers to a machine learning paradigm that combines multiple weak learners to obtain a stronger, more stable, and better generalization-capable ensemble model. By constructing and coordinating multiple basic models, it improves overall prediction accuracy and adaptability to unknown data. The averaging fusion mechanism refers to an ensemble strategy that minimizes the overall prediction error of the ensemble model by weighting or arithmetically averaging the outputs of multiple basic predictors. This is used to synthesize the opinions of various models to obtain a final consensus prediction result and reduce the overfitting risk and variance error of a single model. The adapted quality prediction plugin is constructed based on a number of sample quality predictors and a selection of predictor numbers. Specifically, the selection of predictors is used as the ensemble weight for each sample quality predictor. The current thermal composite process parameters to be predicted are input into the selected predictors to obtain preliminary predicted values ​​for each quality indicator. Then, a weighted average is calculated based on each weight, and the final output of the integrated peel strength, appearance quality coefficient, and barrier performance parameters is the final prediction result of the plugin.

[0053] In summary, the adaptive quality prediction plugin is not a fixed model, but a highly adapted integrated prediction system that is dynamically generated based on the physical properties of the material to be processed. This effectively solves the problem of accurate prediction when historical data for new materials is lacking.

[0054] S40: Based on the aforementioned adaptive quality prediction plugin, with the goal of maximizing the quality of the aluminum-plastic composite strip and the thermal composite process parameter threshold as the optimization space, the thermal composite process parameters are iteratively optimized and searched according to the material property data to obtain the adaptive thermal composite process parameter space.

[0055] Specifically, based on the aforementioned adaptive quality prediction plugin, with the goal of maximizing the quality of the aluminum-plastic composite strip, and using the threshold values ​​of the thermal bonding process parameters as the optimization space, iterative optimization search of the thermal bonding process parameters is performed based on the material property data to obtain the adaptive thermal bonding process parameter space, including:

[0056] Obtain the threshold values ​​of the thermal lamination process parameters of the thermal lamination equipment, including hot pressing temperature, lamination pressure, production line running speed, aluminum foil tension, and plastic film tension.

[0057] Using the aforementioned thermal composite process parameter thresholds, the adaptation quality prediction plugin is tested for predictive stability based on the material property data, and the overall prediction fluctuation coefficient is output.

[0058] Based on the aforementioned quality prediction plugin, with the goal of maximizing the quality of the aluminum-plastic composite strip, and using the threshold values ​​of the thermal composite process parameters as the optimization space, the thermal composite process parameters are iteratively optimized and searched according to the overall prediction fluctuation coefficient and material property data to obtain the adaptive thermal composite process parameter space.

[0059] First, obtain the threshold values ​​for the thermal lamination process parameters of the thermal lamination equipment. Thermal lamination equipment refers to specialized industrial equipment used to laminate aluminum foil and plastic film together by heating and pressurizing. Thermal lamination process parameters refer to the key physical quantities that need to be set and controlled during lamination processing on this equipment, including hot-pressing temperature, lamination pressure, production line speed, aluminum foil tension, and plastic film tension. The threshold values ​​for thermal lamination process parameters refer to the safe upper and lower limits that each process parameter can be set to, including: the hot-pressing temperature threshold, which refers to the lowest and highest safe temperature values ​​that the heating rollers can be set to. This threshold ensures that the hot-melt bonding process of the plastic film can be carried out within its effective flow temperature range, avoiding weak adhesion due to excessively low temperatures or material degradation and charring due to excessively high temperatures; the lamination pressure threshold, which refers to the minimum and maximum pressure values ​​that the pressure roller system can provide. This threshold ensures that the aluminum foil and plastic film can achieve tight bonding under sufficient pressure, while preventing excessive pressure from damaging the materials or the equipment rollers; and the production line speed threshold, which refers to the lowest and highest speed values ​​that the entire production line can operate stably. This threshold directly determines production efficiency and lamination effect. Too slow a speed affects output, while too fast a speed may result in insufficient thermal lamination time, affecting bonding quality. The aluminum foil tension and plastic film tension thresholds refer to the minimum and maximum tension values ​​that the unwinding and rewinding mechanisms can apply to the aluminum foil and plastic film, respectively. This threshold ensures that the material is flat and wrinkle-free during transport. Too little tension will cause the material to wrinkle and drift, while too much tension will cause the aluminum foil to stretch or even break.

[0060] The threshold values ​​for thermal composite process parameters are a quantitative representation of the equipment's capabilities and the safety boundaries of material processing. They are used to define the search space for feasible solutions for subsequent process parameter optimization algorithms, ensuring that all recommended parameters meet the requirements for safe equipment operation and material processing feasibility.

[0061] Furthermore, using the aforementioned thermal composite process parameter thresholds, the adaptation quality prediction plugin is subjected to a prediction stability test based on the material property data, and the overall prediction fluctuation coefficient is output, including:

[0062] To meet the requirement of random distribution of preset parameters, several initial process parameters are randomly selected within the threshold of the thermal composite process parameters.

[0063] The initial process parameters are combined with the material property data to obtain several material processing schemes;

[0064] Randomly select a first material processing scheme, and use the adapted quality prediction plugin to perform Q predictions based on the first material processing scheme, and output Q first quality prediction results;

[0065] Based on the Q first quality prediction results, Q first quality prediction coefficients are evaluated and determined, and prediction fluctuation analysis is performed on the Q first quality prediction coefficients to output a first prediction fluctuation coefficient, wherein the first prediction fluctuation coefficient is the ratio of the standard deviation to the mean of the Q first quality prediction coefficients.

[0066] Several predicted volatility coefficients are obtained through sequential analysis, and the overall predicted volatility coefficient is calculated from the average value.

[0067] First, constrained by the requirement of random distribution of preset parameters, several initial process parameters are randomly selected within the threshold range of thermal composite process parameters. Specifically, within the set threshold range of multiple parameters, several sets of initial process parameters are generated according to a predefined statistical distribution strategy. This distribution strategy can select a space-filling design such as uniform distribution, normal distribution, or Latin hypercube based on prior knowledge to ensure that the sampling points can cover all areas of the entire feasible process space while avoiding local clustering, thereby ensuring the diversity, representativeness, and exploratory nature of the test samples. Each set of generated process parameters strictly meets the constraints of equipment safety and process feasibility, forming the basic input set for subsequent stability testing.

[0068] Secondly, within the threshold range of the thermal bonding process parameters, several initial process parameters are randomly selected. These initial process parameters are then combined with material property data to obtain several material processing schemes. Specifically, the initial process parameters are combined with material property data to ensure that each combination of process parameters is fully associated with the inherent properties of the material under test. This facilitates a comprehensive simulation of the material's potential performance under different processing conditions, and is used to systematically test the response behavior of the quality prediction plugin under different process settings, ultimately resulting in several material processing schemes. Each material processing scheme includes a complete set of thermal bonding process parameter values ​​and the associated dataset of material properties to be processed.

[0069] Furthermore, a first material processing scheme is randomly selected, and the adaptive quality prediction plugin is used to perform Q predictions based on the first material processing scheme, outputting Q first quality prediction results. The first material processing scheme refers to a specific scheme randomly selected from several generated material processing schemes, which includes a set of specific thermal composite process parameter values ​​and corresponding material property data. The first material processing scheme is input into the adaptive quality prediction plugin, and Q predictions are performed to obtain Q sets of predicted values ​​for peel strength, appearance quality coefficient, and barrier performance parameters, which are used as Q first quality prediction results.

[0070] Furthermore, based on the Q first quality prediction results, Q first quality prediction coefficients are determined, and a prediction fluctuation analysis is performed on these Q first quality prediction coefficients to output the first prediction fluctuation coefficient. Here, the Q first quality prediction results are Q sets of quality index data output by the adapted quality prediction plugin after Q independent predictions for the same material processing scheme; the first quality prediction coefficient refers to the numerical sequence extracted from these Q sets of data for a specific quality index, such as peel strength; the first prediction fluctuation coefficient is the ratio of the standard deviation to the mean of the numerical sequence of this quality index, i.e., the coefficient of variation. This coefficient represents the repeatability and stability of the quality prediction plugin's prediction results for this quality index under fixed process and material input conditions. The smaller the value, the more concentrated the prediction results and the more stable the model; the larger the value, the more dispersed the predictions and the more uncertain the model's output at that point. Based on the Q first quality prediction results, Q first quality prediction coefficients are determined, and a prediction fluctuation analysis is performed on these Q first quality prediction coefficients. Prediction fluctuation analysis refers to the process of evaluating the statistical dispersion of multiple prediction outputs under the same input conditions, ultimately yielding the first prediction fluctuation coefficient.

[0071] Finally, the obtained predicted fluctuation coefficients are analyzed sequentially, and the overall predicted fluctuation coefficient is calculated by averaging them. Specifically, the predicted fluctuation coefficients are obtained by repeating the aforementioned test and fluctuation analysis process for different material processing schemes. The overall predicted fluctuation coefficient is the arithmetic mean of these first predicted fluctuation coefficients. By taking the average, the local stability performance of the quality prediction plugin at multiple sampling points throughout the entire thermal composite process parameter space is effectively aggregated, thus obtaining a single quantitative index that can comprehensively evaluate the overall predicted stability of the plugin on the current new material.

[0072] Furthermore, based on the aforementioned adaptive quality prediction plugin, with the goal of maximizing the quality of the aluminum-plastic composite strip and using the threshold values ​​of the thermal bonding process parameters as the optimization space, an iterative optimization search of the thermal bonding process parameters is performed based on the overall predicted fluctuation coefficient and material property data to obtain the adaptive thermal bonding process parameter space, including:

[0073] The ratio of the overall predicted fluctuation coefficient to the preset standard fluctuation coefficient is set as the optimization parameter adjustment coefficient. The initial number of optimization solutions and the initial number of optimization iterations are corrected according to the optimization parameter adjustment coefficient to obtain the number of suitable optimization solutions P and the number of suitable optimization iterations. The initial number of optimization solutions is 20 and the initial number of optimization iterations is 500.

[0074] Obtain the mean values ​​of several quality prediction coefficients corresponding to several initial process parameters, and arrange the several initial process parameters in descending order of the mean values ​​of the quality prediction coefficients to obtain the initial solution sequence;

[0075] The first P solutions of the initial solution sequence are selected as optimal solutions, and the remaining initial solutions are selected as inferior solutions. The inferior solutions are clustered with the P optimal solutions to obtain P solution sets. The number of inferior solutions is at least 20 times the number of optimal solutions. If this condition is not met, the thermal composite process parameter threshold is used to supplement the solution.

[0076] Within the P solution sets, with the optimal solution as the direction, the inferior solutions in the solution set are adjusted according to the preset optimization step size. If the average quality prediction coefficient of the adjusted inferior solution is greater than or equal to the average quality prediction coefficient of the optimal solution, the inferior solution is used to replace the optimal solution, resulting in P updated solution sets.

[0077] Continue iterative optimization until the number of adaptation optimization iterations is reached, output P current optimal solutions from the P current updated solution sets, and add them to the adaptation thermal composite process parameter space.

[0078] First, the ratio of the overall predicted volatility coefficient to the preset standard volatility coefficient is set as the optimization parameter adjustment coefficient. The preset standard volatility coefficient is a baseline volatility level value pre-set based on historical data or domain experience to judge whether the stability of the prediction model is acceptable. The optimization parameter adjustment coefficient = overall predicted volatility coefficient / preset standard volatility coefficient, directly reflecting the uncertainty of the model prediction. A value greater than 1 indicates that the model prediction instability exceeds the baseline, while a value less than 1 indicates that it is better than the baseline, meaning that subsequent optimization algorithms need to adaptively adjust the scaling ratio of the search scale according to this uncertainty. Then, the initial number of optimal solution sets and the initial number of optimization iterations are corrected based on the optimization parameter adjustment coefficient to obtain the number of suitable optimal solution sets P and the number of suitable optimization iterations, i.e., the number of suitable optimal solution sets = the number of initial optimal solution sets × the optimization parameter adjustment coefficient; the number of suitable optimization iterations = the number of initial optimization iterations × the optimization parameter adjustment coefficient. The greater the prediction volatility, i.e., the greater the model uncertainty, the wider the required search range and the longer the search time. The initial number of optimal solution sets is 20, and the initial number of optimization iterations is 500.

[0079] Secondly, the mean values ​​of several quality prediction coefficients corresponding to several initial process parameters are obtained, and the initial process parameters are arranged in descending order of the mean value of the quality prediction coefficients to obtain the initial solution sequence. Specifically, the mean value of the quality prediction coefficient refers to the average value of the quality indicators obtained after each combination of initial process parameters has been predicted multiple times by the adapted quality prediction plugin during the stability test. Arranging the parameters in descending order of the mean value means sorting the process parameter combinations according to their predicted quality from high to low, thus forming the initial solution sequence.

[0080] Furthermore, the first P solutions in the initial solution sequence are selected as optimal solutions, and the remaining initial solutions are designated as inferior solutions. The inferior solutions are then clustered using the P optimal solutions to obtain P solution sets. Specifically, optimal solutions are the top P combinations of process parameters with the highest predicted quality in the initial solution sequence, representing the most promising search direction. Inferior solutions are the remaining lower-quality parameter combinations in the sequence. Using the P optimal solutions as cluster centers, all inferior solutions are assigned to the clusters represented by their closest optimal solutions based on similarity metrics such as Euclidean distance, thus forming P solution sets. To ensure effective clustering and diversity in subsequent optimization, the number of inferior solutions must be at least 20 times the number of optimal solutions. If the actual number of inferior solutions is insufficient, new process parameter combinations must be randomly generated within the threshold range of the thermal composite process parameters to supplement inferior solutions until the required number is met.

[0081] Within P solution sets, inferior solutions are adjusted according to a preset optimization step size, with the optimal solution as the direction. This process is independent and parallel within each solution set. Since the optimal solution is the best solution in its current set, using the optimal solution as the direction means changing the process parameters of the inferior solutions in the direction of the parameter values ​​corresponding to the optimal solution. The preset optimization step size is a positive scalar or vector pre-set based on the parameter magnitude and search accuracy requirements, determining the magnitude of each adjustment. If the average quality prediction coefficient of the adjusted inferior solution is greater than or equal to the average quality prediction coefficient of the optimal solution, then the inferior solution replaces the optimal solution. This indicates that the inferior solution, after adjustment, has become a better or equivalent quality process solution within the current solution set, and the original optimal solution is eliminated, thus completing this iteration and obtaining P updated solution sets.

[0082] The iterative optimization continues until the optimal optimization iteration count is reached. P current optimal solutions from the P currently updated solution sets are then added to the adaptive thermal composite process parameter space. Specifically, this iterative process repeatedly performs directional adjustments to inferior solutions and replacements with superior solutions within the solution sets. Each complete update of all P solution sets is counted as one iteration. When the cumulative iteration count reaches the optimal optimization iteration count, the optimization process terminates. At this point, the current optimal solution stored in each solution set is the P current optimal solution. The process parameter combination corresponding to the optimal solution is taken as the final recommended result and added to the adaptive thermal composite process parameter space. This constitutes a set of high-quality process parameter solutions for the current new material, intelligently optimized and screened, allowing for flexible selection of optimized parameter ranges based on actual production conditions.

[0083] S50: The space of adaptive thermal composite process parameters is used as the recommended process parameters for the material to be processed.

[0084] Finally, the adapted thermal bonding process parameter space is used as the recommended process parameters for the material to be processed. Specifically, this adapted thermal bonding process parameter space is a set of P current optimal solutions, i.e., P sets of high-performance thermal bonding process parameters, ultimately output by an optimization algorithm. All parameters are within the safe threshold range allowed by the equipment and have been predicted and iteratively verified by an intelligent model, significantly improving the overall quality of the aluminum-plastic composite strip prepared from the current material. This recommended process parameter result is not a single parameter value, but a set of high-quality parameter solutions for production applications. Based on the actual equipment status, efficiency requirements, or cost constraints, the most suitable process scheme can be selected for production, thus ensuring product quality while also considering production flexibility and reliability.

[0085] In summary, the embodiments of this application have at least the following technical effects:

[0086] Compared to existing technologies, this application first introduces a physical property similarity coefficient mechanism. Through systematic comparison and similarity measurement of historical material data in the physical property database of new materials and sample materials, it can fully explore and reuse the implicit process rules in existing sample data even in the absence of similar historical data. This overcomes the excessive reliance of traditional data-driven methods on a large amount of similar historical data and significantly improves the algorithm's predictive adaptability and generalization ability for new materials. Secondly, based on the principle of ensemble learning, an adaptive quality prediction plugin is dynamically constructed. It integrates the advantages of multiple basic prediction models and automatically adjusts the model structure and weights according to the specific characteristics of the material to be processed. This achieves a more accurate characterization and prediction of complex nonlinear process-quality relationships, improving the accuracy and adaptability of the quality prediction model.

[0087] Finally, this invention replaces the trial-and-error mode relying on human experience with an intelligent iterative optimization search algorithm. By using the threshold values ​​of thermal composite process parameters as the optimization space and maximizing product quality as the explicit objective function, it automatically performs efficient parameter search and recommendation, reducing the number of trial productions and adjustment cycles. This solves the problems of long development cycles, high resource consumption, and high production costs associated with traditional methods, achieving rapid, accurate, and automated recommendation of aluminum-plastic composite strip production process parameters. While improving product quality stability, it effectively saves production costs.

[0088] Example 2, as Figure 2 As shown, based on the same inventive concept as the intelligent recommendation method for process parameters of aluminum-plastic composite strip provided in Embodiment 1, this embodiment of the invention also provides an intelligent recommendation system for process parameters of aluminum-plastic composite strip, including:

[0089] The material property data acquisition module 11 is used to acquire the material property data of the material to be processed before the aluminum-plastic composite strip is prepared.

[0090] The new material judgment and similarity coefficient calculation module 12 is used to determine whether the material to be processed is a new material based on the sample material property database and the material property data. If it is a new material, it obtains several property similarity coefficients.

[0091] The quality prediction plugin construction module 13 is used to construct an adapted quality prediction plugin based on the principle of ensemble learning, according to the sample quality predictor library and the several physical property similarity coefficients.

[0092] The process parameter optimization search module 14 is used to perform iterative optimization search of the thermal composite process parameters based on the material property data, with the goal of maximizing the quality of the aluminum-plastic composite strip and the thermal composite process parameter threshold as the optimization space, based on the adaptive quality prediction plug-in, to obtain the adaptive thermal composite process parameter space.

[0093] The process parameter recommendation result output module 15 is used to use the adapted thermal composite process parameter space as the process parameter recommendation result of the material to be processed.

[0094] The material property data acquisition module 11 is specifically used for:

[0095] Obtain the materials to be processed for preparing aluminum-plastic composite tape, and the material property data of the materials to be processed. The materials to be processed include aluminum foil, plastic film and adhesive. The material property data of aluminum foil includes at least thickness, density, specific heat capacity, thermal conductivity, surface tension, yield strength and elongation. The material property data of plastic film includes plastic film type and plastic film property information. The material property data of adhesive includes adhesive type and adhesive property information.

[0096] The new material judgment and similarity coefficient calculation module 12 is specifically used for:

[0097] Based on the sample material property database, it is determined whether the material to be processed is a new material according to the material property data. If it is a new material, several property similarity coefficients are obtained, including:

[0098] A sample material property database is constructed based on historical aluminum-plastic composite strip processing records of similar thermal composite equipment, wherein the sample material property database includes several sample material property data.

[0099] The material property data and the material property data of the several samples are compared and similarity is performed to obtain several property similarity coefficients;

[0100] If the number of coefficients with a value of 1 among the plurality of physical property similarity coefficients is not 0, then the material to be processed is an old material; if the number of coefficients with a value of 1 among the plurality of physical property similarity coefficients is 0, then the material to be processed is a new material.

[0101] Specifically, the quality prediction plugin construction module 13 is used for:

[0102] Based on the principle of ensemble learning, an adaptive quality prediction plugin is constructed according to the sample quality predictor library and the aforementioned physical property similarity coefficients, including:

[0103] Based on deep learning, several sample quality predictors are constructed according to the material property data of the several samples, and a sample quality predictor library is obtained. The output of the quality predictor is peel strength, appearance quality coefficient and barrier performance parameter.

[0104] The ratio of the physical property similarity coefficient to the first constant is rounded down to the number of predictors to be selected. Several numbers of predictors to be selected are calculated based on the several physical property similarity coefficients, wherein the first constant is 0.05.

[0105] Based on the principles of ensemble learning and the average fusion mechanism, an adaptive quality prediction plugin is constructed according to the number of sample quality predictors and the number of predictors selected.

[0106] The process parameter optimization search module 14 is specifically used for:

[0107] Based on the aforementioned adaptive quality prediction plugin, with the goal of maximizing the quality of the aluminum-plastic composite strip, and using the threshold values ​​of the thermal bonding process parameters as the optimization space, iterative optimization search of the thermal bonding process parameters is performed based on the material property data to obtain the adaptive thermal bonding process parameter space, including:

[0108] Obtain the threshold values ​​of the thermal lamination process parameters of the thermal lamination equipment, including hot pressing temperature, lamination pressure, production line running speed, aluminum foil tension, and plastic film tension.

[0109] Using the aforementioned thermal composite process parameter thresholds, the adaptation quality prediction plugin is tested for predictive stability based on the material property data, and the overall prediction fluctuation coefficient is output.

[0110] Based on the aforementioned quality prediction plugin, with the goal of maximizing the quality of the aluminum-plastic composite strip, and using the threshold values ​​of the thermal composite process parameters as the optimization space, the thermal composite process parameters are iteratively optimized and searched according to the overall prediction fluctuation coefficient and material property data to obtain the adaptive thermal composite process parameter space.

[0111] Specifically, using the aforementioned thermal composite process parameter thresholds, the adaptation quality prediction plugin is subjected to a prediction stability test based on the material property data, and the overall prediction fluctuation coefficient is output, including:

[0112] To meet the requirement of random distribution of preset parameters, several initial process parameters are randomly selected within the threshold of the thermal composite process parameters.

[0113] The initial process parameters are combined with the material property data to obtain several material processing schemes;

[0114] Randomly select a first material processing scheme, and use the adapted quality prediction plugin to perform Q predictions based on the first material processing scheme, and output Q first quality prediction results;

[0115] Based on the Q first quality prediction results, Q first quality prediction coefficients are evaluated and determined, and prediction fluctuation analysis is performed on the Q first quality prediction coefficients to output a first prediction fluctuation coefficient, wherein the first prediction fluctuation coefficient is the ratio of the standard deviation to the mean of the Q first quality prediction coefficients.

[0116] Several predicted volatility coefficients are obtained through sequential analysis, and the overall predicted volatility coefficient is calculated from the average value.

[0117] Specifically, based on the aforementioned adaptive quality prediction plugin, with the goal of maximizing the quality of the aluminum-plastic composite strip, and using the threshold values ​​of the thermal bonding process parameters as the optimization space, the thermal bonding process parameters are iteratively optimized and searched based on the overall predicted fluctuation coefficient and material property data to obtain the adaptive thermal bonding process parameter space, including:

[0118] The ratio of the overall predicted fluctuation coefficient to the preset standard fluctuation coefficient is set as the optimization parameter adjustment coefficient. The initial number of optimization solutions and the initial number of optimization iterations are corrected according to the optimization parameter adjustment coefficient to obtain the number of suitable optimization solutions P and the number of suitable optimization iterations. The initial number of optimization solutions is 20 and the initial number of optimization iterations is 500.

[0119] Obtain the mean values ​​of several quality prediction coefficients corresponding to several initial process parameters, and arrange the several initial process parameters in descending order of the mean values ​​of the quality prediction coefficients to obtain the initial solution sequence;

[0120] The first P solutions of the initial solution sequence are selected as optimal solutions, and the remaining initial solutions are selected as inferior solutions. The inferior solutions are clustered with the P optimal solutions to obtain P solution sets. The number of inferior solutions is at least 20 times the number of optimal solutions. If this condition is not met, the thermal composite process parameter threshold is used to supplement the solution.

[0121] Within the P solution sets, with the optimal solution as the direction, the inferior solutions in the solution set are adjusted according to the preset optimization step size. If the average quality prediction coefficient of the adjusted inferior solution is greater than or equal to the average quality prediction coefficient of the optimal solution, the inferior solution is used to replace the optimal solution, resulting in P updated solution sets.

[0122] Continue iterative optimization until the number of adaptation optimization iterations is reached, output P current optimal solutions from the P current updated solution sets, and add them to the adaptation thermal composite process parameter space.

[0123] The process parameter recommendation result output module 15 is specifically used for:

[0124] It receives and processes the spatial data of adaptive thermal composite process parameters from the process parameter optimization search module, converts it into a standardized set of process parameter instructions that can be directly recognized and executed by the production system, and displays and outputs the instruction set and the corresponding visual recommendation scheme through the human-machine interface.

[0125] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0126] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0127] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. An aluminum-plastic composite tape process parameter intelligent recommendation method, characterized in that, The method comprises: Before the preparation of the aluminum-plastic composite tape, the material physical property data of the material to be processed is obtained; Based on the historical aluminum-plastic composite tape processing records of similar thermal compounding equipment, a sample material physical property database is constructed, based on the sample material physical property database, whether the material to be processed is a new material is judged according to the material physical property data, if it is a new material, a plurality of physical property similarity coefficients are obtained, wherein the sample material physical property database comprises a plurality of sample material physical property data; Based on the principle of ensemble learning, the adaptive quality prediction plug-in is constructed according to the sample quality predictor library and the plurality of physical property similarity coefficients; Based on the adaptive quality prediction plug-in, the iterative optimization search of the thermal compounding process parameters is carried out according to the material physical property data, with the maximum aluminum-plastic composite tape quality as the target and the thermal compounding process parameter threshold as the optimization space, to obtain the adaptive thermal compounding process parameter space; The adaptive thermal compounding process parameter space is taken as the process parameter recommendation result of the material to be processed; Based on the principle of ensemble learning, the adaptive quality prediction plug-in is constructed according to the sample quality predictor library and the plurality of physical property similarity coefficients, comprising: Based on deep learning, a plurality of sample quality predictors are constructed according to the plurality of sample material physical property data, respectively, to obtain a sample quality predictor library, wherein the sample material physical property data and different thermal compounding process parameters are taken as input features, and the actual measured peel strength, appearance quality coefficient and barrier performance parameters corresponding to the sample under a specific process are taken as training labels, a plurality of deep neural network models are independently trained by supervised learning as a plurality of sample quality predictors; The ratio of the physical property similarity coefficient to the first constant is rounded to the predictor selection quantity, and the plurality of predictor selection quantities are calculated according to the plurality of physical property similarity coefficients, wherein the first constant is 0.05; Based on the principle of ensemble learning and the average fusion mechanism, the adaptive quality prediction plug-in is constructed according to the plurality of sample quality predictors and the plurality of predictor selection quantities. 2.The aluminum-plastic composite tape process parameter intelligent recommendation method according to claim 1, characterized in that, The material to be processed for preparing the aluminum-plastic composite tape is obtained, and the material physical property data of the material to be processed is obtained, wherein the material to be processed comprises aluminum foil, plastic film and adhesive, the material physical property data of the aluminum foil at least comprises thickness, density, specific heat capacity, thermal conductivity, surface tension, yield strength and elongation, the material physical property data of the plastic film comprises plastic film type and plastic film attribute information, and the material physical property data of the adhesive comprises adhesive type and adhesive attribute information. 3.The aluminum-plastic composite tape process parameter intelligent recommendation method according to claim 1, characterized in that, Based on the sample material physical property database, whether the material to be processed is a new material is judged according to the material physical property data, if it is a new material, a plurality of physical property similarity coefficients are obtained, comprising: The material physical property data and the plurality of sample material physical property data are respectively subjected to similarity traversal comparison to obtain a plurality of physical property similarity coefficients; If the number of coefficients equal to 1 in the plurality of physical property similarity coefficients is not 0, the material to be processed is an old material, and if the number of coefficients equal to 1 in the plurality of physical property similarity coefficients is 0, the material to be processed is a new material. 4.The aluminum-plastic composite tape process parameter intelligent recommendation method according to claim 1, characterized in that, Based on the adaptive quality prediction plug-in, taking maximizing the quality of the aluminum-plastic composite tape as a target, taking the hot compounding process parameter threshold as an optimization space, and according to the material physical property data, iterative optimization search is performed on the hot compounding process parameter to obtain an adaptive hot compounding process parameter space, including: Obtaining a hot compounding process parameter threshold of a hot compounding equipment, wherein the hot compounding process parameter includes hot pressing temperature, compounding pressure, production line running speed, aluminum foil tension and plastic film tension; Using the hot compounding process parameter threshold, performing a prediction stability test on the adaptive quality prediction plug-in according to the material physical property data, and outputting an overall prediction fluctuation coefficient; Based on the adaptive quality prediction plug-in, taking maximizing the quality of the aluminum-plastic composite tape as a target, taking the hot compounding process parameter threshold as an optimization space, and according to the overall prediction fluctuation coefficient and the material physical property data, iterative optimization search is performed on the hot compounding process parameter to obtain an adaptive hot compounding process parameter space. 5.The aluminum-plastic composite tape process parameter intelligent recommendation method according to claim 4, characterized in that, Using the hot compounding process parameter threshold, performing a prediction stability test on the adaptive quality prediction plug-in according to the material physical property data, and outputting an overall prediction fluctuation coefficient, including: Randomly selecting a plurality of initial process parameters within the hot compounding process parameter threshold, with the constraint of meeting the preset parameter random distribution requirement; Combining the plurality of initial process parameters with the material physical property data to obtain a plurality of material processing schemes; Randomly selecting a first material processing scheme, using the adaptive quality prediction plug-in, and performing Q times of prediction according to the first material processing scheme to output Q first quality prediction results; According to the Q first quality prediction results, determining Q first quality prediction coefficients, and performing prediction fluctuation analysis on the Q first quality prediction coefficients to output a first prediction fluctuation coefficient, wherein the first prediction fluctuation coefficient is the ratio of the standard deviation to the mean of the Q first quality prediction coefficients; A plurality of prediction fluctuation coefficients are obtained by sequential analysis, and the overall prediction fluctuation coefficient is calculated by averaging. 6.The aluminum-plastic composite tape process parameter intelligent recommendation method according to claim 5, characterized in that, Based on the adaptive quality prediction plug-in, taking maximizing the quality of the aluminum-plastic composite tape as a target, taking the hot compounding process parameter threshold as an optimization space, and according to the overall prediction fluctuation coefficient and the material physical property data, iterative optimization search is performed on the hot compounding process parameter to obtain an adaptive hot compounding process parameter space, including: Setting the ratio of the overall prediction fluctuation coefficient to a preset standard fluctuation coefficient as an optimization parameter adjustment coefficient, correcting the initial optimization solution set number and the initial optimization iteration number according to the optimization parameter adjustment coefficient to obtain an adaptive optimization solution set number P and an adaptive optimization iteration number, wherein the initial optimization solution set number is 20 and the initial optimization iteration number is 500; Obtaining a plurality of quality prediction coefficient means corresponding to a plurality of initial process parameters, and arranging the plurality of initial process parameters in descending order of quality prediction coefficient mean to obtain an initial solution sequence; Select the first P solutions of the initial solution sequence as optimal solutions, and the remaining initial solutions as inferior solutions, and cluster the inferior solutions with the P optimal solutions to obtain P solution sets, wherein the number of inferior solutions is at least 20 times the number of optimal solutions, if not, use the thermal compounding process parameter threshold to supplement; Within the P solution sets, adjust the inferior solutions in the solution set according to the preset optimization step length with the optimal solution as the direction, if the quality prediction coefficient mean of the adjusted inferior solution is greater than or equal to the quality prediction coefficient mean of the optimal solution, replace the optimal solution with the inferior solution to obtain P updated solution sets; Continue to iterate and optimize until the adaptive optimization iteration number is reached, output the P current optimal solutions in the P current updated solution sets, and add them to the adaptive thermal compounding process parameter space.

7. An aluminum-plastic composite tape process parameter intelligent recommendation system, characterized in that, For performing the method of any one of claims 1-6, comprising: a material property data acquisition module for acquiring material property data of a material to be processed before the preparation of the aluminum-plastic composite tape; a new material judgment and similarity coefficient calculation module for constructing a sample material property database based on historical aluminum-plastic composite tape processing records of similar thermal compounding equipment, and judging whether the material to be processed is a new material based on the sample material property database according to the material property data, if it is a new material, obtaining a plurality of material property similarity coefficients, wherein the sample material property database includes a plurality of sample material property data; a quality prediction plug-in construction module for constructing an adaptive quality prediction plug-in based on the ensemble learning principle according to a sample quality predictor library and the plurality of material property similarity coefficients; a process parameter optimization search module for performing iterative optimization search of thermal compounding process parameters based on the adaptive quality prediction plug-in, with maximizing the quality of aluminum-plastic composite tape as the target and the thermal compounding process parameter threshold as the optimization space, to obtain an adaptive thermal compounding process parameter space; a process parameter recommendation result output module for outputting the adaptive thermal compounding process parameter space as the process parameter recommendation result of the material to be processed.

Images