Preparation method of high-heat-efficiency polyurethane heating film based on generative adversarial network optimization

Abstract

The invention discloses a preparation method of a high-heat-efficiency polyurethane heating film based on generative adversarial network optimization. The preparation method comprises the following steps: S1, screening and characterizing a polyurethane base material, a conductive filler and an interface enhancer; s2, generating an initial data set based on a cross-domain mapping rule; s3, constructing a cross-domain attribute migration generative adversarial network, and outputting an optimized formula parameter set; s4, generating an optimal material formula in combination with a dynamic attribute weight adjustment mechanism; s5, preparing a homogeneous mixture through ultrasonic dispersion and shear stirring technologies; s6, optimizing a channel structure by utilizing a collaborative functional material nano-channel reconstruction module and combining a dynamic interface self-assembly technology to generate a to-be-formed film; s7, the film to be formed is tested, and the material formula and the channel structure are optimized; and S8, forming of the thin film is completed through a coating and film forming process, and finally the optimized polyurethane heating film is prepared. By means of the generative adversarial network optimization technology, efficient preparation of the polyurethane heating film is achieved.

Description

Technical Field

[0001] The invention relates to the technical field of polyurethane heating film preparation, and in particular to a method for preparing a high-thermal-efficiency polyurethane heating film based on generative adversarial network optimization. Background Art

[0002] As an efficient, flexible and customizable thermal conductive material, polyurethane heating film has been widely used in flexible electronic devices, medical heating films, car seat heating and other fields. Its core technology involves the improvement of thermal conductivity, the optimization of mechanical properties and the control of manufacturing processes. The preparation method of traditional polyurethane heating film mainly relies on simple material mixing and hot pressing molding process. Although the basic performance requirements have been achieved to a certain extent, its performance improvement is subject to many factors, such as the dispersibility of conductive fillers, the interface bonding strength between substrates and fillers, and the design of thermal conductive network structures, and it has always been difficult to break through the performance bottleneck.

[0003] In the prior art, the mainstream method is to improve the thermal conductivity of polyurethane heating films by introducing conductive fillers such as graphene, carbon nanotubes, and silver nanowires. However, the dispersion of conductive fillers and the continuity of thermal conduction paths are difficult to accurately control in traditional processes. The distribution of fillers in polyurethane substrates is often random, resulting in high thermal resistance, making it difficult to fully utilize the thermal conductivity potential of fillers. At the same time, the interface thermal resistance between fillers is large, which also seriously limits the improvement of thermal conductivity. In addition, due to the high concentration of fillers, the mechanical flexibility and toughness of the film are often sacrificed, causing the material to be easy to break and not resistant to bending, thus affecting its actual application effect.

[0004] In the existing preparation methods, there is always a contradiction between the proportion of conductive fillers added and the mechanical properties of the polyurethane substrate. When the proportion of conductive fillers is increased to improve thermal conductivity, the flexibility and mechanical strength of the film decrease significantly; conversely, when the flexibility of the substrate is improved, the thermal conductivity is difficult to meet actual needs. This type of single-objective optimization method often cannot take into account the balance of multi-objective performance. In addition, the traditional preparation process lacks the ability to design the thermal conductive network structure, and the randomly distributed fillers cannot form an efficient heat flow channel, resulting in thermal conductivity far lower than the theoretical calculated value.

[0005] With the rapid development of artificial intelligence technology, data-driven methods such as generative adversarial networks (GANs) have gradually been applied in the field of material design. In the existing technology, some studies have attempted to use GANs to predict material properties, but most of them are limited to the optimization of a single goal, such as only improving thermal conductivity, ignoring the constraints of mechanical properties and manufacturing processes. In addition, the application of GANs is mostly based on a single data source, such as experimental data or simulation data, and fails to effectively integrate cross-domain data from experimental tests and physical field simulations, thus limiting the prediction accuracy and scope of application of the model.

[0006] In terms of preparation process, traditional coating and film-forming technology relies on empirical methods to control key parameters (such as coating thickness, drying temperature and hot pressing pressure), and lacks model-based optimization methods. This empirical method easily leads to instability and poor repeatability of process parameters, which adversely affects the consistency of film performance. In addition, the parameter selection during hot pressing and curing often cannot match the material formulation and thermal channel design, further reducing the overall performance of the film.

[0007] Therefore, the existing technology has the following major defects: first, the conductive filler is not sufficiently dispersed in the polyurethane substrate, resulting in limited improvement in thermal conductivity; second, there are multi-objective contradictions in the optimization of material properties, and it is difficult to simultaneously balance thermal conductivity, mechanical properties and manufacturing process; third, the thermal conductive network structure lacks precise design capabilities and cannot effectively utilize the thermal conductivity potential of the filler; finally, the parameter control of the coating and film-forming process relies on empirical methods, which makes it difficult to achieve precise optimization of the entire process.

[0008] Therefore, how to provide a method for preparing a high-thermal-efficiency polyurethane heating film based on generative adversarial network optimization is an urgent problem that technicians in this field need to solve. Summary of the invention

[0009] One purpose of the present invention is to propose a method for preparing a high thermal efficiency polyurethane heating film based on generative adversarial network optimization. The present invention utilizes generative adversarial network optimization technology to achieve the formula design, thermal network optimization and preparation process control of a high thermal efficiency polyurethane heating film by integrating experimental data and physical field simulation data. This method successfully constructs a branched nano-thermal conductive network, improves the continuity and uniformity of the heat flow inside the film, and at the same time, through dynamic optimization of the coating and film-forming process, achieves the coordinated optimization of the film's high thermal conductivity, high mechanical strength and good flexibility. The present invention has the significant advantages of excellent performance, high preparation efficiency and strong process controllability, and provides important technical support for the research and development and application of high-performance thin film materials.

[0010] According to an embodiment of the present invention, a method for preparing a high thermal efficiency polyurethane heating film based on generative adversarial network optimization comprises the following steps:

[0011] S1. Screen and characterize polyurethane substrates, conductive fillers and interface enhancers to obtain the thermal conductivity, resistivity, mechanical strength and interface bonding parameters of the materials;

[0012] S2, obtaining experimental data and simulation data through experimental testing and physical field simulation, and performing standardization processing, and generating an initial data set based on cross-domain mapping rules;

[0013] S3. Construct a cross-domain attribute transfer generative adversarial network, separate the material formula parameters in the initial data set into physical effects and chemical effects through the attribute separator in the generator for independent encoding, and introduce a cross-domain adversarial module in the discriminator to output an optimized formula parameter set;

[0014] S4. Based on the optimized formulation parameter set and combined with the dynamic property weight adjustment mechanism, the conductive filler ratio, interface enhancer concentration and film thickness are optimized to generate the optimal material formulation;

[0015] S5, performing a mixing pretreatment on the generated optimal material formula, uniformly dispersing the polyurethane substrate, the conductive filler and the interface enhancer by ultrasonic dispersion and shear stirring technology to prepare a homogeneous mixture;

[0016] S6. Based on the homogeneous mixture, using the synergistic functional material nanochannel reconstruction module, inducing the conductive filler to form a branched nano-thermal conductive network along a specific path through a template method, and optimizing the channel structure in combination with dynamic interface self-assembly technology to generate a film to be formed with an optimized thermal conductive network;

[0017] S7, performing thermal conductivity, resistivity and mechanical property tests on the film to be formed, inputting the test results as feedback data into the multi-objective inversion optimization network in the cross-domain attribute transfer generative adversarial network, dynamically adjusting the parameters of the generator and the discriminator, and further optimizing the material formula and channel structure;

[0018] S8. Based on the optimized material formula and channel structure, the film is formed through coating and film-forming processes, and the coating thickness, drying temperature, hot pressing pressure and curing time are controlled during the curing process to finally prepare an optimized polyurethane heating film.

[0019] Optionally, the S3 specifically includes:

[0020] S31, constructing a cross-domain attribute transfer generative adversarial network, initializing a generator and a discriminator, wherein the generator is used to input material formula parameters and independently encode the parameters for physical effects and chemical effects through an attribute separator, and the discriminator is used to dynamically evaluate the physical credibility and chemical rationality of the generated data;

[0021] S32, using a property separator in the generator, dividing the initial data set and the material formula parameters into physical effect parameters and chemical effect parameters, wherein the physical effect parameters include thermal conductivity, resistivity and heat flow distribution, and the chemical effect parameters include interface bonding force and filler compatibility;

[0022] S33. Based on the physical effect parameters, establish a generation objective function and a discriminator objective function in the physical field, and define the physical effect optimization objective function of the generator:

[0023]

[0024] Among them, L phy (G) represents the physical effect optimization objective function of the generator, x represents the real data sample, represents the distribution of real physical effect data, D phy (x) represents the scoring function of the discriminator for the real physical effect data, z represents the noise vector, p z (z) represents the distribution of the noise vector, G(z) represents the physical effect optimization result generated by the generator, and P phy (G(z)) represents the predicted value of the physical properties of the generated data, P phy (x real ) represents the performance value of the real physical effect data, λ1 represents the weight coefficient of the second norm error, λ2 represents the weight coefficient of the variance constraint, Represents the square error of the generated result and the true value, Var(P phy (G(z)) represents the variance of the physical effect value of the generated data;

[0025] S34. Based on the chemical effect parameters, establish the generation objective function and discriminator objective function in the chemical field, and define the chemical effect optimization objective function of the generator:

[0026]

[0027] Among them, L chem (G) represents the chemical effect optimization objective function of the generator, represents the distribution of real chemical effect data, D chem (x) represents the scoring function of the discriminator for the real chemical effect data, C chem (G(z)) represents the chemical property prediction value of the generated data, μ1 represents the weight coefficient of Manhattan distance, μ2 represents the weight coefficient of relative entropy, C chem (x real ) represents the performance value of the real chemical effect data, ∥·∥1 represents the Manhattan distance between the generated result and the real value, and KL(·||·) represents the KL relative entropy between the generated data and the real data distribution;

[0028] S35. Introduce a cross-domain adversarial module in the discriminator, combine physical credibility evaluation and chemical rationality evaluation, and generate cross-domain optimization results by dynamically adjusting the weights of the physical and chemical fields;

[0029] S36. By generating an objective function through a dynamic optimization generator and combining the optimization results of the physical effect parameters and the chemical effect parameters, an optimized formula parameter set is output, wherein the optimized formula parameter set includes a conductive filler ratio, an interface enhancer concentration, and a film thickness.

[0030] Optionally, the S4 specifically includes:

[0031] S41, extracting conductive filler ratio, interface enhancer concentration and film thickness parameters from the optimized formula parameter set as initial inputs for dynamic adjustment, and defining the parameter range of the initial inputs;

[0032] S42. Build a dynamic property weight adjustment mechanism to set initial weights according to the target performance requirements of the film, where the physical property weights are used to optimize thermal conductivity and resistivity, the chemical property weights are used to enhance interface compatibility and dispersibility, and the manufacturing process weights are used to control film thickness and uniformity;

[0033] S43. Optimize the proportion of conductive fillers and calculate the effect of filler proportion on thermal conductivity and resistivity according to the dynamic property weight adjustment mechanism:

[0034]

[0035] in, represents the optimized conductive filler ratio, represents the initial conductive filler ratio, η represents the learning rate, and F phy Represents the loss function of physical performance, F chem represents the loss function of chemical properties, F manu represents the loss function of the manufacturing process, C f represents the proportion of conductive filler, α represents the weight of physical properties, β represents the weight of chemical properties, and γ represents the weight of manufacturing process;

[0036] S44. Optimize the concentration of the interfacial enhancer, analyze the interfacial bonding force and filler dispersion according to the optimization target, and dynamically adjust the enhancer concentration to maximize the interfacial compatibility;

[0037] S45. In combination with the uniformity requirements of the manufacturing process for the film thickness, the film thickness parameters are adjusted on the basis of the optimized conductive filler ratio and the reinforcing agent concentration; the optimization of the film thickness is based on the stability of the scraping process, while taking into account the uniformity of the curing conditions in the film forming process and the performance of the finished product;

[0038] S46. The optimal material formula is generated by comprehensively optimizing the conductive filler ratio, interface enhancer concentration and film thickness parameters.

[0039] Optionally, the S6 specifically includes:

[0040] S61, based on the initial material state of the homogeneous mixture, applying the synergistic functional material nanochannel reconstruction module to the layout of the conductive filler, and constructing an initial thermal conductive network by a template method;

[0041] S62. Based on the initial thermal conductive network, the dynamic interface self-assembly technology is used to adjust the interaction force between the conductive filler and the polyurethane substrate through the interface enhancer, induce the rearrangement of the filler in the microstructure, and optimize the distribution uniformity of the filler. The dynamic interface self-assembly technology relies on the active groups on the surface of the filler to construct a thermal conductive network in the polyurethane substrate with a low resistance path;

[0042] S63. By adjusting the branch network structure of the heat conduction channel, the number of branches of the heat conduction path is increased, and the connection method between the branches and the trunk is optimized. The branch network structure is centered on multi-level branch nodes, and the contact area of ​​the conductive filler is increased to reduce the interface thermal resistance and improve the overall thermal conductivity.

[0043] S64, combining the synergistic functional material nanochannel reconstruction module, adjusting the microstructure parameters of the heat conduction network, wherein the microstructure parameters include branch density, branch width and number of nodes, while reducing local overheating;

[0044] S65. The bonding force between the conductive filler and the polyurethane substrate is enhanced by the chemical reaction of the interface enhancer, and the interfacial thermal resistance between the conductive fillers is reduced by using the nano-modifier; the dynamic interface self-assembly technology is combined with the curing process of the polyurethane substrate to stabilize the heat conduction channel structure inside the polyurethane substrate;

[0045] S66. Finally, a film to be formed having an optimized thermal conductive network is formed. The thermal conductive network inside the film to be formed is mainly composed of branched nanochannels, and the branch paths are clear and evenly distributed.

[0046] Optionally, the S7 specifically includes:

[0047] S71, performing a thermal conductivity test on the film to be formed, measuring the thermal conductivity value of the film by a laser flash method, recording the test data and comparing it with the target performance index; evaluating the structural effectiveness of the heat conduction channel and the distribution uniformity of the conductive filler by analyzing the deviation of the test results;

[0048] S72, testing the resistivity of the film to be formed, measuring the resistivity of the film by a four-probe method, recording the difference between the test value and the target resistivity, and analyzing the electrical conductivity of the film in combination with the microscopic characteristics of the distribution of the conductive filler, and identifying the local area of ​​excessively high resistance in the thermal conductive network;

[0049] S73, testing the mechanical properties of the film to be formed, measuring the tensile strength, flexibility modulus and elongation at break of the film by a tensile test method, recording the data and comparing them with the target mechanical performance indicators, and identifying the factors affecting the mechanical properties by evaluating the influence of the heat conduction channel on the overall mechanical properties of the substrate;

[0050] S74, using the test results of thermal conductivity, resistivity and mechanical properties as input data, and performing comprehensive analysis through a multi-objective inversion optimization network; the multi-objective inversion optimization network uses each performance indicator as an independent optimization target, establishes a correlation model between the performances, analyzes the potential trade-off relationship between the target performances, and generates global feedback on the optimization direction;

[0051] S75. In a multi-objective inverse optimization network, the input parameters are adjusted by the generator, and the performance evaluation criteria of the discriminator are redefined. The parameter weights of the generator and the discriminator are dynamically adjusted. The heat conduction channel structure is optimized by combining multi-source data, so that the material formula generated by the network meets the target requirements of thermal conductivity, resistivity and mechanical properties at the same time.

[0052] S76. Based on the feedback results of the multi-objective inversion optimization network, an updated material formula and channel structure are generated to form a new design of the film to be formed. The thermal conductivity, resistivity and mechanical properties of the optimized film are retested and optimized repeatedly.

[0053] Optionally, the S8 specifically includes:

[0054] S81, based on the optimized material formula and channel structure, prepare the coating material, introduce the uniformly mixed material into the coating device, and set initial parameters, wherein the initial parameters include the coating thickness, coating speed and coating pressure;

[0055] S82. Use the scraping process to evenly coat the material on the surface of the polyurethane substrate. Use the scraping equipment to monitor the coating thickness in real time, and dynamically adjust the scraping parameters based on the optimization goals:

[0056]

[0057] Where T represents the scraping thickness after real-time adjustment, T0 represents the initial scraping thickness, and δ vis Indicates the rate of change of material viscosity, v rel Indicates the relative deviation of scraping speed from the optimal speed, μ eff Indicates the effective viscosity of the material;

[0058] S83, after coating, the film is sent to a drying device, the drying temperature range is set to 80°C to 120°C, and the drying temperature and time are dynamically adjusted in combination with a generative adversarial network. During the drying process, the solvent evaporation rate on the surface and inside of the film is controlled by hot air flow;

[0059] S84. Performing hot pressing on the dried film, setting hot pressing conditions and optimizing hot pressing pressure:

[0060]

[0061] Where P represents the actual hot pressing pressure, P opt represents the optimal hot pressing pressure, σ res represents the residual stress inside the film, E represents the elastic modulus of the film material, and t press Indicates the actual hot pressing time, t opt Indicates the optimal heat pressing time;

[0062] S85, curing the hot-pressed film, controlling the curing time and temperature gradient, gradually reducing the temperature from high temperature to room temperature, and optimizing the bonding force between the conductive filler and the substrate in combination with dynamic interface reaction;

[0063] S86. Perform performance tests on the formed polyurethane heating film to ensure that the thickness uniformity, thermal conductivity, resistivity and mechanical properties of the film are within the preset target range, and finally produce an optimized polyurethane heating film.

[0064] The beneficial effects of the present invention are:

[0065] First, the present invention introduces a cross-domain attribute transfer generative adversarial network, integrates experimental data and physical field simulation data, and constructs a performance prediction and optimization model. This cross-domain data fusion breaks through the limitation of single data in the prior art and greatly improves the accuracy and reliability of material performance optimization. At the same time, by utilizing the attribute separation mechanism in the generator and the dynamic evaluation capability of the discriminator, the complex trade-offs between thermal conductivity, mechanical properties and manufacturing process of the material are accurately optimized.

[0066] Secondly, the present invention has achieved innovation in the design of thermal conductive networks. By coordinating functional material nanochannel reconstruction modules and dynamic interface self-assembly technology, a branched nano-thermal conductive network has been successfully constructed. This thermal conductive network is centered on multi-level branch nodes, the distribution of fillers in the substrate is more uniform, the continuity of the heat flow path is significantly improved, and the thermal resistance inside the film is significantly reduced. At the same time, the introduction of the interface enhancer optimizes the bonding force between the conductive filler and the substrate, which not only improves the thermal conductivity, but also ensures the flexibility and mechanical strength of the film. Compared with the traditional random distribution of fillers, the present invention solves the problems of high thermal resistance and uneven performance in traditional technologies through precise network design.

[0067] In addition, the present invention has significant advantages in optimizing the preparation process. By generating dynamic optimization parameters provided by the adversarial network, the coating thickness, drying temperature, hot pressing pressure and curing time in the coating and film-forming process are precisely controlled, avoiding the parameter instability caused by relying on empirical methods in traditional processes. Especially in the hot pressing and curing stages, the present invention optimizes the structural stability of the thermal conductive network by dynamically adjusting the pressure and temperature, and effectively avoids the concentration of stress in the film through temperature gradient control, further improving the uniformity and comprehensive performance of the film. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0069] Figure 1 This is an overall flow chart of the method for preparing a high thermal efficiency polyurethane heating film based on generative adversarial network optimization proposed in the present invention;

[0070] Figure 2 This is a schematic diagram of the cross-domain attribute transfer generative adversarial network architecture of the high thermal efficiency polyurethane heating film preparation method based on generative adversarial network optimization proposed in the present invention. DETAILED DESCRIPTION

[0071] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0072] refer to Figure 1 and Figure 2 , a method for preparing a high thermal efficiency polyurethane heating film based on generative adversarial network optimization comprises the following steps:

[0073] S1. Screen and characterize polyurethane substrates, conductive fillers and interface enhancers to obtain the thermal conductivity, resistivity, mechanical strength and interface bonding parameters of the materials;

[0074] S2, obtaining experimental data and simulation data through experimental testing and physical field simulation, and performing standardization processing, and generating an initial data set based on cross-domain mapping rules;

[0075] S3. Construct a cross-domain attribute transfer generative adversarial network, separate the material formula parameters in the initial data set into physical effects and chemical effects through the attribute separator in the generator for independent encoding, and introduce a cross-domain adversarial module in the discriminator to output an optimized formula parameter set;

[0076] S4. Based on the optimized formulation parameter set and combined with the dynamic property weight adjustment mechanism, the conductive filler ratio, interface enhancer concentration and film thickness are optimized to generate the optimal material formulation;

[0077] S5, performing a mixing pretreatment on the generated optimal material formula, uniformly dispersing the polyurethane substrate, the conductive filler and the interface enhancer by ultrasonic dispersion and shear stirring technology to prepare a homogeneous mixture;

[0078] S6. Based on the homogeneous mixture, using the synergistic functional material nanochannel reconstruction module, inducing the conductive filler to form a branched nano-thermal conductive network along a specific path through a template method, and optimizing the channel structure in combination with dynamic interface self-assembly technology to generate a film to be formed with an optimized thermal conductive network;

[0079] S7, performing thermal conductivity, resistivity and mechanical property tests on the film to be formed, inputting the test results as feedback data into the multi-objective inversion optimization network in the cross-domain attribute transfer generative adversarial network, dynamically adjusting the parameters of the generator and the discriminator, and further optimizing the material formula and channel structure;

[0080] S8. Based on the optimized material formula and channel structure, the film is formed through coating and film-forming processes, and the coating thickness, drying temperature, hot pressing pressure and curing time are controlled during the curing process to finally prepare an optimized polyurethane heating film.

[0081] In this implementation, S3 specifically includes:

[0082] S31, constructing a cross-domain attribute transfer generative adversarial network, initializing a generator and a discriminator, wherein the generator is used to input material formula parameters and independently encode the parameters for physical effects and chemical effects through an attribute separator, and the discriminator is used to dynamically evaluate the physical credibility and chemical rationality of the generated data;

[0083] S32, using a property separator in the generator, dividing the initial data set and the material formula parameters into physical effect parameters and chemical effect parameters, wherein the physical effect parameters include thermal conductivity, resistivity and heat flow distribution, and the chemical effect parameters include interface bonding force and filler compatibility;

[0084] S33. Based on the physical effect parameters, establish a generation objective function and a discriminator objective function in the physical field, and define the physical effect optimization objective function of the generator:

[0085]

[0086] Among them, L phy (G) represents the physical effect optimization objective function of the generator, x represents the real data sample, represents the distribution of real physical effect data, D phy (x) represents the scoring function of the discriminator for the real physical effect data, z represents the noise vector, p z (z) represents the distribution of the noise vector, G(z) represents the physical effect optimization result generated by the generator, and P phy (G(z)) represents the predicted value of the physical properties of the generated data, P phy (x real ) represents the performance value of the real physical effect data, λ1 represents the weight coefficient of the second norm error, λ2 represents the weight coefficient of the variance constraint, Represents the square error of the generated result and the true value, Var(P phy (G(z)) represents the variance of the physical effect value of the generated data;

[0087] S34. Based on the chemical effect parameters, establish the generation objective function and discriminator objective function in the chemical field, and define the chemical effect optimization objective function of the generator:

[0088]

[0089] Among them, L chem (G) represents the chemical effect optimization objective function of the generator, represents the distribution of real chemical effect data, D chem (x) represents the scoring function of the discriminator for the real chemical effect data, C chem (G(z)) represents the chemical property prediction value of the generated data, μ1 represents the weight coefficient of Manhattan distance, μ2 represents the weight coefficient of relative entropy, C chem (x real ) represents the performance value of the real chemical effect data, ||·||1 represents the Manhattan distance between the generated result and the real value, and KL(·||·) represents the KL relative entropy between the generated data and the real data distribution;

[0090] S35. Introduce a cross-domain adversarial module in the discriminator, combine physical credibility evaluation and chemical rationality evaluation, and generate cross-domain optimization results by dynamically adjusting the weights of the physical and chemical fields;

[0091] S36. By generating an objective function through a dynamic optimization generator and combining the optimization results of the physical effect parameters and the chemical effect parameters, an optimized formula parameter set is output, wherein the optimized formula parameter set includes a conductive filler ratio, an interface enhancer concentration, and a film thickness.

[0092] In this implementation, S4 specifically includes:

[0093] S41, extracting conductive filler ratio, interface enhancer concentration and film thickness parameters from the optimized formula parameter set as initial inputs for dynamic adjustment, and defining the parameter range of the initial inputs;

[0094] S42. Build a dynamic property weight adjustment mechanism to set initial weights according to the target performance requirements of the film, where the physical property weights are used to optimize thermal conductivity and resistivity, the chemical property weights are used to enhance interface compatibility and dispersibility, and the manufacturing process weights are used to control film thickness and uniformity;

[0095] S43. Optimize the proportion of conductive fillers and calculate the effect of filler proportion on thermal conductivity and resistivity according to the dynamic property weight adjustment mechanism:

[0096]

[0097] in, represents the optimized conductive filler ratio, represents the initial conductive filler ratio, η represents the learning rate, and F phy Represents the loss function of physical performance, F chem represents the loss function of chemical properties, F manu represents the loss function of the manufacturing process, C f represents the proportion of conductive filler, α represents the weight of physical properties, β represents the weight of chemical properties, and γ represents the weight of manufacturing process;

[0098] S44. Optimize the concentration of the interfacial enhancer, analyze the interfacial bonding force and filler dispersion according to the optimization target, and dynamically adjust the enhancer concentration to maximize the interfacial compatibility;

[0099] S45. In combination with the uniformity requirements of the manufacturing process for the film thickness, the film thickness parameters are adjusted on the basis of the optimized conductive filler ratio and the reinforcing agent concentration; the optimization of the film thickness is based on the stability of the scraping process, while taking into account the uniformity of the curing conditions in the film forming process and the performance of the finished product;

[0100] S46. The optimal material formula is generated by comprehensively optimizing the conductive filler ratio, interface enhancer concentration and film thickness parameters.

[0101] In this implementation manner, S6 specifically includes:

[0102] S61, based on the initial material state of the homogeneous mixture, applying the synergistic functional material nanochannel reconstruction module to the layout of the conductive filler, and constructing an initial thermal conductive network by a template method;

[0103] S62. Based on the initial thermal conductive network, the dynamic interface self-assembly technology is used to adjust the interaction force between the conductive filler and the polyurethane substrate through the interface enhancer, induce the rearrangement of the filler in the microstructure, and optimize the distribution uniformity of the filler. The dynamic interface self-assembly technology relies on the active groups on the surface of the filler to construct a thermal conductive network in the polyurethane substrate with a low resistance path;

[0104] S63. By adjusting the branch network structure of the heat conduction channel, the number of branches of the heat conduction path is increased, and the connection method between the branches and the trunk is optimized. The branch network structure is centered on multi-level branch nodes, and the contact area of ​​the conductive filler is increased to reduce the interface thermal resistance and improve the overall thermal conductivity.

[0105] S64, combining the synergistic functional material nanochannel reconstruction module, adjusting the microstructure parameters of the heat conduction network, wherein the microstructure parameters include branch density, branch width and number of nodes, while reducing local overheating;

[0106] S65. The bonding force between the conductive filler and the polyurethane substrate is enhanced by the chemical reaction of the interface enhancer, and the interfacial thermal resistance between the conductive fillers is reduced by using the nano-modifier; the dynamic interface self-assembly technology is combined with the curing process of the polyurethane substrate to stabilize the heat conduction channel structure inside the polyurethane substrate;

[0107] S66. Finally, a film to be formed having an optimized thermal conductive network is formed. The thermal conductive network inside the film to be formed is mainly composed of branched nanochannels, and the branch paths are clear and evenly distributed.

[0108] In this implementation manner, the S7 specifically includes:

[0109] S71, performing a thermal conductivity test on the film to be formed, measuring the thermal conductivity value of the film by a laser flash method, recording the test data and comparing it with the target performance index; evaluating the structural effectiveness of the heat conduction channel and the distribution uniformity of the conductive filler by analyzing the deviation of the test results;

[0110] S72, testing the resistivity of the film to be formed, measuring the resistivity of the film by a four-probe method, recording the difference between the test value and the target resistivity, and analyzing the electrical conductivity of the film in combination with the microscopic characteristics of the distribution of the conductive filler, and identifying the local area of ​​excessively high resistance in the thermal conductive network;

[0111] S73, testing the mechanical properties of the film to be formed, measuring the tensile strength, flexibility modulus and elongation at break of the film by a tensile test method, recording the data and comparing them with the target mechanical performance indicators, and identifying the factors affecting the mechanical properties by evaluating the influence of the heat conduction channel on the overall mechanical properties of the substrate;

[0112] S74, using the test results of thermal conductivity, resistivity and mechanical properties as input data, and performing comprehensive analysis through a multi-objective inversion optimization network; the multi-objective inversion optimization network uses each performance indicator as an independent optimization target, establishes a correlation model between the performances, analyzes the potential trade-off relationship between the target performances, and generates global feedback on the optimization direction;

[0113] S75. In a multi-objective inverse optimization network, the input parameters are adjusted by the generator, and the performance evaluation criteria of the discriminator are redefined. The parameter weights of the generator and the discriminator are dynamically adjusted. The heat conduction channel structure is optimized by combining multi-source data, so that the material formula generated by the network meets the target requirements of thermal conductivity, resistivity and mechanical properties at the same time.

[0114] S76. Based on the feedback results of the multi-objective inversion optimization network, an updated material formula and channel structure are generated to form a new design of the film to be formed. The thermal conductivity, resistivity and mechanical properties of the optimized film are retested and optimized repeatedly.

[0115] In this implementation manner, S8 specifically includes:

[0116] S81, based on the optimized material formula and channel structure, prepare the coating material, introduce the uniformly mixed material into the coating device, and set initial parameters, wherein the initial parameters include the coating thickness, coating speed and coating pressure;

[0117] S82. Use the scraping process to evenly coat the material on the surface of the polyurethane substrate. Use the scraping equipment to monitor the coating thickness in real time, and dynamically adjust the scraping parameters based on the optimization goals:

[0118]

[0119] Where T represents the scraping thickness after real-time adjustment, T0 represents the initial scraping thickness, and δ vis Indicates the rate of change of material viscosity, v rel Indicates the relative deviation of the scraping speed from the optimal speed, μ eff Indicates the effective viscosity of the material;

[0120] S83, after coating, the film is sent to a drying device, the drying temperature range is set to 80°C to 120°C, and the drying temperature and time are dynamically adjusted in combination with a generative adversarial network. During the drying process, the solvent evaporation rate on the surface and inside of the film is controlled by hot air flow;

[0121] S84. Performing hot pressing on the dried film, setting hot pressing conditions and optimizing hot pressing pressure:

[0122]

[0123] Where P represents the actual hot pressing pressure, P opt represents the optimal hot pressing pressure, σ res represents the residual stress inside the film, E represents the elastic modulus of the film material, and t press Indicates the actual hot pressing time, t opt Indicates the optimal heat pressing time;

[0124] S85, curing the hot-pressed film, controlling the curing time and temperature gradient, gradually reducing the temperature from high temperature to room temperature, and optimizing the bonding force between the conductive filler and the substrate in combination with dynamic interface reaction;

[0125] S86. Perform performance tests on the formed polyurethane heating film to ensure that the thickness uniformity, thermal conductivity, resistivity and mechanical properties of the film are within the preset target range, and finally produce an optimized polyurethane heating film.

[0126] Embodiment 1:

[0127] In order to verify the feasibility of the present invention in implementation, the present invention is applied to a certain enterprise, which needs to develop an efficient heat dissipation film for wearable smart bracelets. The internal electronic components of the bracelet generate high heat under high-intensity usage scenarios (such as continuous heart rate monitoring, GPS recording, etc.). Traditional heat dissipation films cannot meet the comfort requirements of long-term wear, and the film's insufficient mechanical properties make it easy to break. The present invention optimizes the proportion of conductive fillers, the concentration of interface enhancers and the thickness of the film, combines coating and film forming process control, prepares a polyurethane heating film with excellent performance, and conducts performance testing.

[0128] In this embodiment, the preparation method of the present invention uses graphene as a conductive filler and a silane coupling agent as an interface enhancer. The initial parameters are 10% graphene, 2% silane coupling agent concentration, and 30 microns film thickness. The material formula is optimized using a cross-domain attribute migration generative adversarial network, and the optimal formula parameters are finally determined to be: 12.5% ​​graphene, 3.2% silane coupling agent concentration, and 25 microns film thickness. The optimized heating film has higher thermal conductivity, lower resistivity, and excellent mechanical properties.

[0129] When applied in flexible electronic devices, the laboratory simulated the high-load operating environment of the bracelet: the device power was 1.2W and the ambient temperature was 25°C. The test results show that after the polyurethane heating film of the present invention worked continuously for 4 hours, the device surface temperature was maintained below 40°C, while the device surface temperature of the traditional heat dissipation film reached above 60°C under the same conditions. In addition, the heating film of the present invention was repeatedly bent 1000 times under the condition of a bending radius of less than 10mm, and its mechanical properties did not change significantly, the tensile strength remained above 15MPa, and the elongation at break was higher than 150%.

[0130] Table 1 Performance comparison between polyurethane heating film and traditional heat dissipation film

[0131]

[0132]

[0133] Table 2 Test data of heat dissipation efficiency of different films under high load conditions

[0134] Time (hours) Surface temperature of the heating film of the present invention (℃) Surface temperature of traditional heat dissipation film (℃) 1 36 54 2 38 58 3 39 59 4 40 60

[0135] Table 3 Mechanical performance durability test data of the heating film of the present invention

[0136] Bending radius(mm) Bending times Tensile strength retention rate (%) Retention rate of elongation at break (%) 10 500 98 97 10 1000 95 93 5 500 90 89 5 1000 88 85

[0137] It can be seen from the experimental data comparison table that the polyurethane heating film prepared by the present invention has significant advantages in performance over traditional heat dissipation films. According to the data in Table 1 above, the thermal conductivity of the present invention reaches 1.8W / m·K, which is 63.6% higher than the 1.1W / m·K of the traditional heat dissipation film. This significant improvement is due to the design of the branched nano-thermal conductive network and the uniform distribution of the conductive filler, which optimizes the heat flow conduction path of the film. In addition, the resistivity of the present invention is 4.5Ω·cm, which is 37.5% lower than the 7.2Ω·cm of the traditional heat dissipation film, indicating that the conductive properties of the film have been significantly improved, which is of great significance for the rapid release of heat energy and the conduction of electrical energy in practical applications.

[0138] In terms of mechanical properties, the tensile strength of the heating film of the present invention reaches 15.2MPa, which is 60% higher than the 9.5MPa of the traditional heat dissipation film, and the elongation at break is also increased from 95% to 153%, an increase of 61.1%. These data show that the present invention not only has a breakthrough improvement in thermal conductivity, but also solves the problem of poor mechanical flexibility of traditional heat dissipation films, showing stronger tensile and bending resistance.

[0139] The heat dissipation efficiency test under high load conditions listed in Table 2 above further verifies the superiority of the present invention. After 4 hours of continuous operation, the heating film of the present invention can stably control the surface temperature of the equipment below 40°C, while the traditional heat dissipation film reaches 60°C, a difference of up to 20°C. This shows that the high thermal efficiency polyurethane heating film of the present invention significantly improves the heat dissipation capacity in practical applications, ensuring the safety of the equipment under high-intensity operation and the comfort of the user.

[0140] It can be seen from the mechanical property durability test data in Table 3 above that when the bending radius of the polyurethane heating film of the present invention is 10mm, even after 1000 repeated bendings, its tensile strength can still be maintained at 95%, and the elongation at break can be maintained at 93%. Under the condition of a smaller bending radius (5mm), after 1000 tests, the tensile strength and elongation at break are also maintained at 88% and 85% respectively. These data show that the heating film of the present invention has extremely high durability in the application scenario of flexible equipment, and the mechanical properties can be maintained at a high level even under complex use conditions.

[0141] Through the analysis of this implementation, it can be found that the present invention has successfully solved the problems of low thermal conductivity, poor mechanical properties and insufficient durability of traditional heat dissipation films by comprehensively optimizing thermal conductivity, mechanical properties and heat dissipation efficiency. In the practical application of flexible electronic devices, the polyurethane heating film of the present invention not only significantly improves the heat dissipation efficiency of the equipment, but also prolongs the service life of the film, providing important technical support for the design and promotion of high-performance heat dissipation films.

[0142] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for preparing a high thermal efficiency polyurethane heating film based on generative adversarial network optimization, characterized in that: The steps include: S1. Screen and characterize polyurethane substrates, conductive fillers and interface enhancers to obtain the thermal conductivity, resistivity, mechanical strength and interface bonding parameters of the materials; S2, obtaining experimental data and simulation data through experimental testing and physical field simulation, and performing standardization processing, and generating an initial data set based on cross-domain mapping rules; S3. Construct a cross-domain attribute transfer generative adversarial network, separate the material formula parameters in the initial data set into physical effects and chemical effects through the attribute separator in the generator for independent encoding, and introduce a cross-domain adversarial module in the discriminator to output an optimized formula parameter set; S4. Based on the optimized formulation parameter set and combined with the dynamic property weight adjustment mechanism, the conductive filler ratio, interface enhancer concentration and film thickness are optimized to generate the optimal material formulation; S5, performing a mixing pretreatment on the generated optimal material formula, uniformly dispersing the polyurethane substrate, the conductive filler and the interface enhancer by ultrasonic dispersion and shear stirring technology to prepare a homogeneous mixture; S6. Based on the homogeneous mixture, using the synergistic functional material nanochannel reconstruction module, inducing the conductive filler to form a branched nano-thermal conductive network along a specific path through a template method, and optimizing the channel structure in combination with dynamic interface self-assembly technology to generate a film to be formed with an optimized thermal conductive network; S7, performing thermal conductivity, resistivity and mechanical property tests on the film to be formed, inputting the test results as feedback data into the multi-objective inversion optimization network in the cross-domain attribute transfer generative adversarial network, dynamically adjusting the parameters of the generator and the discriminator, and further optimizing the material formula and channel structure; S8. Based on the optimized material formula and channel structure, the film is formed through coating and film-forming processes, and the coating thickness, drying temperature, hot pressing pressure and curing time are controlled during the curing process to finally prepare an optimized polyurethane heating film.

2. The method for preparing a high thermal efficiency polyurethane heating film based on generative adversarial network optimization according to claim 1 is characterized in that: The S3 specifically includes: S31, constructing a cross-domain attribute transfer generative adversarial network, initializing a generator and a discriminator, wherein the generator is used to input material formula parameters and independently encode the parameters for physical effects and chemical effects through an attribute separator, and the discriminator is used to dynamically evaluate the physical credibility and chemical rationality of the generated data; S32, using a property separator in the generator, dividing the initial data set and the material formula parameters into physical effect parameters and chemical effect parameters, wherein the physical effect parameters include thermal conductivity, resistivity and heat flow distribution, and the chemical effect parameters include interface bonding force and filler compatibility; S33. Based on the physical effect parameters, establish a generation objective function and a discriminator objective function in the physical field, and define the physical effect optimization objective function of the generator: Among them, L phy (G) represents the physical effect optimization objective function of the generator, x represents the real data sample, represents the distribution of real physical effect data, D phy (x) represents the scoring function of the discriminator for the real physical effect data, z represents the noise vector, p z (z) represents the distribution of the noise vector, G(z) represents the physical effect optimization result generated by the generator, and P phy (G(z)) represents the predicted value of the physical properties of the generated data, P phy (x real ) represents the performance value of the real physical effect data, λ1 represents the weight coefficient of the second norm error, λ2 represents the weight coefficient of the variance constraint, Represents the square error of the generated result and the true value, Var(P phy (G(z)) represents the variance of the physical effect value of the generated data; S34. Based on the chemical effect parameters, establish the generation objective function and discriminator objective function in the chemical field, and define the chemical effect optimization objective function of the generator: Among them, L chem (G) represents the chemical effect optimization objective function of the generator, represents the distribution of real chemical effect data, D chem (x) represents the scoring function of the discriminator for the real chemical effect data, C chem (G(z)) represents the chemical property prediction value of the generated data, μ1 represents the weight coefficient of Manhattan distance, μ2 represents the weight coefficient of relative entropy, C chem (x real ) represents the performance value of the real chemical effect data, ||·||1 represents the Manhattan distance between the generated result and the real value, and KL(·||·) represents the KL relative entropy between the generated data and the real data distribution; S35. Introduce a cross-domain adversarial module in the discriminator, combine physical credibility evaluation and chemical rationality evaluation, and generate cross-domain optimization results by dynamically adjusting the weights of the physical and chemical fields; S36. By generating an objective function through a dynamic optimization generator and combining the optimization results of the physical effect parameters and the chemical effect parameters, an optimized formula parameter set is output, wherein the optimized formula parameter set includes a conductive filler ratio, an interface enhancer concentration, and a film thickness.

3. The method for preparing a high thermal efficiency polyurethane heating film based on generative adversarial network optimization according to claim 1 is characterized in that: The S4 specifically includes: S41, extracting conductive filler ratio, interface enhancer concentration and film thickness parameters from the optimized formula parameter set as initial inputs for dynamic adjustment, and defining the parameter range of the initial inputs; S42. Build a dynamic property weight adjustment mechanism to set initial weights according to the target performance requirements of the film, where the physical property weights are used to optimize thermal conductivity and resistivity, the chemical property weights are used to enhance interface compatibility and dispersibility, and the manufacturing process weights are used to control film thickness and uniformity; S43. Optimize the proportion of conductive fillers and calculate the effect of filler proportion on thermal conductivity and resistivity according to the dynamic property weight adjustment mechanism: in, represents the optimized conductive filler ratio, represents the initial conductive filler ratio, η represents the learning rate, and F phy Represents the loss function of physical performance, F chem represents the loss function of chemical properties, F manu represents the loss function of the manufacturing process, C f represents the proportion of conductive filler, α represents the weight of physical properties, β represents the weight of chemical properties, and γ represents the weight of manufacturing process; S44. Optimize the concentration of the interfacial enhancer, analyze the interfacial bonding force and filler dispersion according to the optimization target, and dynamically adjust the enhancer concentration to maximize the interfacial compatibility; S45. In combination with the uniformity requirements of the manufacturing process for the film thickness, the film thickness parameters are adjusted on the basis of the optimized conductive filler ratio and the reinforcing agent concentration; the optimization of the film thickness is based on the stability of the scraping process, while taking into account the uniformity of the curing conditions in the film forming process and the performance of the finished product; S46. The optimal material formula is generated by comprehensively optimizing the conductive filler ratio, interface enhancer concentration and film thickness parameters.

4. The method for preparing a high thermal efficiency polyurethane heating film based on generative adversarial network optimization according to claim 1 is characterized in that: The S6 specifically includes: S61, based on the initial material state of the homogeneous mixture, applying the synergistic functional material nanochannel reconstruction module to the layout of the conductive filler, and constructing an initial thermal conductive network by a template method; S62. Based on the initial thermal conductive network, the dynamic interface self-assembly technology is used to adjust the interaction force between the conductive filler and the polyurethane substrate through the interface enhancer, induce the rearrangement of the filler in the microstructure, and optimize the distribution uniformity of the filler. The dynamic interface self-assembly technology relies on the active groups on the surface of the filler to construct a thermal conductive network in the polyurethane substrate with a low resistance path; S63. By adjusting the branch network structure of the heat conduction channel, the number of branches of the heat conduction path is increased, and the connection method between the branches and the trunk is optimized. The branch network structure is centered on multi-level branch nodes, and the contact area of ​​the conductive filler is increased to reduce the interface thermal resistance and improve the overall thermal conductivity. S64, combining the synergistic functional material nanochannel reconstruction module, adjusting the microstructure parameters of the heat conduction network, wherein the microstructure parameters include branch density, branch width and number of nodes, while reducing local overheating; S65. The bonding force between the conductive filler and the polyurethane substrate is enhanced by the chemical reaction of the interface enhancer, and the interfacial thermal resistance between the conductive fillers is reduced by using the nano-modifier; the dynamic interface self-assembly technology is combined with the curing process of the polyurethane substrate to stabilize the heat conduction channel structure inside the polyurethane substrate; S66. Finally, a film to be formed having an optimized thermal conductive network is formed. The thermal conductive network inside the film to be formed is mainly composed of branched nanochannels, and the branch paths are clear and evenly distributed.

5. The method for preparing a high thermal efficiency polyurethane heating film based on generative adversarial network optimization according to claim 1 is characterized in that: The S7 specifically includes: S71, performing a thermal conductivity test on the film to be formed, measuring the thermal conductivity value of the film by a laser flash method, recording the test data and comparing it with the target performance index; evaluating the structural effectiveness of the heat conduction channel and the distribution uniformity of the conductive filler by analyzing the deviation of the test results; S72, testing the resistivity of the film to be formed, measuring the resistivity of the film by a four-probe method, recording the difference between the test value and the target resistivity, and analyzing the electrical conductivity of the film in combination with the microscopic characteristics of the distribution of the conductive filler, and identifying the local area of ​​excessively high resistance in the thermal conductive network; S73, testing the mechanical properties of the film to be formed, measuring the tensile strength, flexibility modulus and elongation at break of the film by a tensile test method, recording the data and comparing them with the target mechanical performance indicators, and identifying the factors affecting the mechanical properties by evaluating the influence of the heat conduction channel on the overall mechanical properties of the substrate; S74, using the test results of thermal conductivity, resistivity and mechanical properties as input data, and performing comprehensive analysis through a multi-objective inversion optimization network; the multi-objective inversion optimization network uses each performance indicator as an independent optimization target, establishes a correlation model between the performances, analyzes the potential trade-off relationship between the target performances, and generates global feedback on the optimization direction; S75. In a multi-objective inverse optimization network, the input parameters are adjusted by the generator, and the performance evaluation criteria of the discriminator are redefined. The parameter weights of the generator and the discriminator are dynamically adjusted. The heat conduction channel structure is optimized by combining multi-source data, so that the material formula generated by the network meets the target requirements of thermal conductivity, resistivity and mechanical properties at the same time. S76. Based on the feedback results of the multi-objective inversion optimization network, an updated material formula and channel structure are generated to form a new design of the film to be formed. The thermal conductivity, resistivity and mechanical properties of the optimized film are retested and optimized repeatedly.

6. The method for preparing a high thermal efficiency polyurethane heating film based on generative adversarial network optimization according to claim 1 is characterized in that: The S8 specifically includes: S81, based on the optimized material formula and channel structure, prepare the coating material, introduce the uniformly mixed material into the coating device, and set initial parameters, wherein the initial parameters include the coating thickness, coating speed and coating pressure; S82. Use the scraping process to evenly coat the material on the surface of the polyurethane substrate. Use the scraping equipment to monitor the coating thickness in real time, and dynamically adjust the scraping parameters based on the optimization goals: Where T represents the scraping thickness after real-time adjustment, T0 represents the initial scraping thickness, and δ vis Indicates the rate of change of material viscosity, v rel Indicates the relative deviation of the scraping speed from the optimal speed, μ eff Indicates the effective viscosity of the material; S83, after coating, the film is sent to a drying device, the drying temperature range is set to 80°C to 120°C, and the drying temperature and time are dynamically adjusted in combination with a generative adversarial network. During the drying process, the solvent evaporation rate on the surface and inside of the film is controlled by hot air flow; S84. Perform hot pressing on the dried film, set hot pressing conditions and optimize hot pressing pressure: Where P represents the actual hot pressing pressure, P opt represents the optimal hot pressing pressure, σ res represents the residual stress inside the film, E represents the elastic modulus of the film material, and t press Indicates the actual hot pressing time, t opt Indicates the optimal heat pressing time; S85, curing the hot-pressed film, controlling the curing time and temperature gradient, gradually reducing the temperature from high temperature to room temperature, and optimizing the bonding force between the conductive filler and the substrate in combination with dynamic interface reaction; S86. Perform performance tests on the formed polyurethane heating film to ensure that the thickness uniformity, thermal conductivity, resistivity and mechanical properties of the film are within the preset target range, and finally produce an optimized polyurethane heating film.

Images