Development method of machine learning-assisted high-thermal-conductivity polymer composite material

Abstract

The invention belongs to the technical field of high-molecular materials, and discloses a machine learning assisted high-thermal-conductivity high-molecular composite material development method, which comprises the following steps: S1, establishing a feature database; s2, data preprocessing; s3, constructing a machine learning model; s4, optimizing the prediction model; s5, screening the prediction model; s6, outputting a prediction result; and S7, verifying a prediction result. Based on machine learning and prediction of the components and the preparation process of the high-thermal-conductivity polymer composite material, the feature database and the machine learning model are fused, development of the high-thermal-conductivity intelligent polymer composite material can be accelerated, the research and development period can be shortened, high-temperature equipment is not needed, and the preparation cost is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of polymer materials, and in particular to a method for developing a high thermal conductivity polymer composite material assisted by machine learning. Background Art

[0002] As microelectronic devices are moving towards miniaturization and high integration, thermal management is particularly critical. In this context, thermally conductive polymer composites have become an ideal choice for solving thermal management problems due to their advantages such as lightness, easy processing and cost-effectiveness, especially in the fields of 5G communications, electronic packaging and energy transmission. However, the actual thermal conductivity of these composites often does not meet expectations, which reveals that the understanding of their heat conduction mechanism is still insufficient. In addition, the long R&D cycle and high cost of materials are also the main bottlenecks restricting their development.

[0003] Carbon fiber, with its graphite-like layered structure that is highly oriented along the fiber axis, not only provides ultra-high strength and rigidity, but also gives the material a certain degree of flexibility and toughness through weak van der Waals forces. Carbon fiber has an extremely low coefficient of thermal expansion, which makes it exhibit excellent dimensional stability when the temperature changes. Although the thermal conductivity of carbon fiber is not as good as that of some metals, it still has high thermal conductivity among non-metallic materials, especially along the fiber direction. In addition, the performance stability of carbon fiber in high temperature environment and its chemical stability to most acids, bases and solvents make it the material of choice for many high-end applications.

[0004] Similar to carbon fiber, hexagonal boron nitride (h-BN) also plays an important role in high-performance applications due to its layered structure, excellent thermal stability, electrical insulation and good thermal conductivity. The weak interaction between its layers gives it excellent lubricity, making h-BN have broad application prospects in high-temperature lubrication, insulating coatings and advanced ceramic composites.

[0005] In the prior art, the Chinese invention patent application with application number 202410175505.5 submitted by Liaoning University discloses a photoresponsive aerogel phase change composite material and its preparation method. The preparation method provided by the invention includes: ultrasonically treating the photoresponsive material and boron nitride to form a photoresponsive material / boron nitride dispersion; then adding the polymer solution to the above dispersion to form a uniform and stable mixed solution, pouring the mixed solution into a mold for vacuum freeze drying to obtain a photoresponsive aerogel material; finally, the photoresponsive aerogel is compounded with the molten mixture of the photoresponsive material and the phase change material by vacuum assisted adsorption. The invention relates to a photoresponsive material, and the research and development of the material is similar to the traditional method, and does not involve application fields such as the research and development of machine learning auxiliary materials.

[0006] The Chinese invention patent application with application number 202110062307.4 submitted by North University of China discloses "Graphene oxide aerogel-based electromagnetic shielding polymer composite material with conductive-thermal double network structure and preparation method thereof". The invention relates to a graphene functional composite material, specifically a graphene oxide aerogel-based electromagnetic shielding polymer composite material with conductive-thermal double network structure and a preparation method thereof. The preparation method comprises: freeze-drying and high-temperature heat treating the graphene oxide solution to obtain a highly conductive three-dimensional graphene aerogel; evenly mixing boron nitride and silicone rubber and filling them into the highly conductive three-dimensional graphene aerogel to obtain a graphene oxide aerogel-based electromagnetic shielding polymer composite material. The invention requires high-temperature equipment and has high energy consumption during the preparation process. Summary of the invention

[0007] In view of the deficiencies of the above-mentioned prior art, the object of the present invention is to provide a method for developing a high thermal conductivity polymer composite material assisted by machine learning. The present invention is based on machine learning and prediction of the components and preparation process of the high thermal conductivity polymer composite material, and integrates the feature database and machine learning model. It can accelerate the development of high thermal conductivity intelligent polymer composite materials, shorten the research and development cycle, and does not require high-temperature equipment, thereby reducing preparation costs.

[0008] In order to achieve the above object, the present invention provides the following technical solutions: A method for developing a high thermal conductivity polymer composite material assisted by machine learning, comprising the following steps: S1: Building a feature database According to the basic process steps for preparing high thermal conductivity polymer composite materials and the characteristic data related to the target performance, the relevant data on the components, scales, proportions, processes and conditions for preparing high thermal conductivity polymer composite materials are collected to establish a characteristic database; S2: Data preprocessing The feature data is converted into binary sub-category features "0" and "1" using one hot encoding to obtain a feature data set; the thermal conductivity of the high thermal conductivity polymer composite material is used as the target variable, and the features in the feature data set are used as independent variables to preprocess the feature data and delete redundant features; the preprocessed feature data set is then divided, with 85% of the feature data set used as a training set and the remaining 15% as a test set; S3: Building Machine Learning Models A recursive elimination nested Ridge regression algorithm model based on a machine learning framework is constructed as a machine learning model, and the pre-processed data is used for training to obtain a pre-trained machine learning model as a prediction model; S4: Optimizing the prediction model Five-fold cross validation was used to evaluate the model, optimize parameters, and obtain the optimized Ridge regression prediction model; S5: Prediction model screening The target performance parameters of the high thermal conductivity polymer composite material to be obtained, including thermal conductivity, are input into the prediction model, which performs feature engineering screening on the relevant feature data prepared; specifically, the SHAP nested Ridge regression prediction model based on the machine learning framework is first used for calculation, and then the positive and negative correlation and importance between the independent variable and the target variable are analyzed, and the calculation results of the regression prediction model are explained; S6: Output prediction results According to the prediction results explained by the regression prediction model, the preparation process of high thermal conductivity polymer composite materials is designed and the prediction results are output; S7: Verify the prediction results According to the prediction results, the high thermal conductivity polymer composite material is actually prepared to obtain the composite material, and then the thermal conductivity performance is tested and verified. The test values ​​are compared with the predicted values ​​to verify the accuracy of the model prediction results.

[0009] The target performance-related characteristic data in step S1 include: thermoplastic polymer material components, filler components and content ratios, molding methods, and additives for preparing composite materials; the fillers include carbon fibers and boron nitride.

[0010] The carbon fiber is a polydopamine-coated carbon fiber, which is prepared by the following steps: B1: Ultrasonic washing of carbon fiber in solvent to remove surface impurities, then complete drying in an oven, and then polydopamine coating of the washed carbon fiber; B2: Polydopamine coating process: The pH value of the Tris buffer solution was adjusted to less than 8.5, and then carbon fiber and dopamine were added to the mixed solution in different proportions and reacted at 30-60°C for 12-36 h; B3: The treated carbon fiber is then placed in a beaker and repeatedly washed in deionized water to obtain polydopamine-coated carbon fiber, which is then dried in an oven for later use.

[0011] The boron nitride is hydroxylated boron nitride, which is prepared by the following steps: C1: Boron nitride was hydroxylated as follows: 1.5-3 g of h-BN was added to 100-200 mL of a mixed solution (isopropanol and deionized water) and subjected to tip sonication. C2: The obtained dispersion is then centrifuged at a speed of 500-1000 for 5-20 min, the obtained upper suspension is centrifuged at a speed of 6000-12000 for 5-20 min, and then dried to obtain hydroxylated boron nitride for use.

[0012] The basic process steps for preparing the high thermal conductivity polymer composite material in step S1 are: A1: Set the mass ratio range of the main raw materials. The proportion of thermoplastic polymer material ranges from 20-90wt%; carbon fiber 1-30wt%, boron nitride 1-30wt%, additive 1-30wt%; A2: Setting the main preparation process The molding preparation processes include: unidirectional freezing, bidirectional freezing, array freezing, and melt blending.

[0013] The thermoplastic polymer material includes: polyethylene glycol, polyurethane, polycaprolactone, paraffin, thermoplastic polyimide; The fillers are carbon fiber and boron nitride; The auxiliary agents include: a dispersant and a covalent cross-linking agent; The dispersant is a mixture of one or more of deionized water and ethanol; and the covalent cross-linking agent is a mixture of one or more of 1,4-butanediol diglycidyl ether and epichlorohydrin.

[0014] In the step A2, when the melt blending molding method is selected, the filler, the additive and the thermoplastic polymer material are first melt blended at high temperature and the bubbles are removed; then the mixture is placed in a room temperature environment and naturally cooled to solidify the thermoplastic polymer to obtain a high thermal conductivity polymer composite material.

[0015] The step A2, when the one-way freezing molding method is selected for preparation, comprises the following steps: D1: Hydroxylated boron nitride is prepared into a dispersion with a concentration of 10-120 mg / mL, and an acid is used as a catalyst to react with a covalent crosslinking agent for 20-60 min to produce a chemical crosslinking reaction; then the polydopamine-coated carbon fiber is added to the dispersant to prepare a uniformly mixed dispersion; D2: Pour the well-mixed dispersion into a plastic beaker and place it on a cold table immersed in liquid nitrogen for directional freezing.

[0016] D3: Treat with a freeze dryer for 12-96 h and post-treat for 1-6 h to obtain boron nitride / carbon fiber aerogel.

[0017] D4: The thermoplastic polymer material is melted and degassed at high temperature, and vacuum impregnated to fill the voids of the aerogel; it is then placed in a room temperature environment and naturally cooled to solidify the thermoplastic polymer to obtain a high thermal conductivity polymer composite material.

[0018] The step A2, when the bidirectional freezing molding method is selected for preparation, comprises the following steps: E1: Hydroxylated boron nitride is prepared into a dispersion with a concentration of 10-120 mg / mL, and an acid is used as a catalyst to react with a covalent crosslinking agent for 20-60 min to produce a chemical crosslinking reaction; then, the polydopamine-coated carbon fiber is added to the dispersant to prepare a uniformly mixed dispersion; E2: The uniformly mixed dispersion was poured into a custom-made wedge-shaped mold (the inclination angle of the wedge-shaped mold was 15°) and placed on a cold stage immersed in liquid nitrogen for directional freezing.

[0019] E3: Treat with a freeze dryer for 12-96 h and post-treat for 1-6 h to obtain boron nitride / carbon fiber aerogel.

[0020] E4: The thermoplastic polymer material is melted and degassed at high temperature, and vacuum impregnated to fill the voids of the aerogel; it is then placed in a room temperature environment and naturally cooled to solidify the thermoplastic polymer to obtain a high thermal conductivity polymer composite material.

[0021] The step A2, when selecting the array freezing molding method for preparation, comprises the following steps: F1: Hydroxylated boron nitride is prepared into a dispersion with a concentration of 10-120 mg / mL, and an acid is used as a catalyst to react with a covalent crosslinking agent for 20-60 min to produce a chemical crosslinking reaction; then the polydopamine-coated carbon fiber is added to the dispersant to prepare a uniformly mixed dispersion; F2: The well-mixed dispersion was poured into a customized circular copper tube array mold and placed on a cold stage immersed in liquid nitrogen for directional freezing; F3: treated with freeze dryer for 12-96 h and post-treated for 1-6 h to obtain boron nitride / carbon fiber aerogel; F4: The thermoplastic polymer material is melted and degassed at high temperature, and vacuum impregnated to fill the voids of the aerogel; it is then placed in a room temperature environment and naturally cooled to solidify the thermoplastic polymer to obtain a high thermal conductivity polymer composite material.

[0022] The beneficial effects of the present invention are: 1. The method for developing high thermal conductivity polymer composite materials assisted by machine learning provided by the present invention is based on machine learning and prediction of the components and preparation process of high thermal conductivity polymer composite materials, and integrates feature database and machine learning model. It can accelerate the development of high thermal conductivity intelligent polymer composite materials, shorten the research and development cycle, and does not require high-temperature equipment, reducing preparation costs.

[0023] 2. The key point of this invention is that by using the Ridge regression algorithm, the influence of various factors on the thermal conductivity of composite materials is analyzed, and the influence of key features on target performance is evaluated. Different composite materials are designed, which greatly shortens the design and development cycle of materials and significantly improves R&D efficiency. Through data-driven methods, a wider range of material combinations and structural configurations can be explored to accelerate the development of innovative materials.

[0024] The above is an overview of the technical solution of the invention. The present invention will be further described below in conjunction with specific implementation methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 A schematic diagram of the process of machine learning in the method for developing the present invention; Figure 2 This is a schematic diagram of the screening result of recursively eliminating nested Ridge features according to an embodiment of the present invention; Figure 3 A schematic diagram of the results of a five-fold cross validation and an independent test set of the Ridge regression model for predicting thermal conductivity of a composite material according to an embodiment of the present invention; Figure 4 It is a schematic diagram of feature importance of the SHAP nested Ridge regression model according to an embodiment of the present invention; Figure 5 Schematic diagram of the positive and negative relationship between the features and targets of the SHAP nested Ridge regression model according to an embodiment of the present invention. DETAILED DESCRIPTION

[0026] The technical solution of the present invention is further described in detail below in conjunction with the embodiments, but the protection scope of the present invention is not limited thereto.

[0027] Example 1 The method for developing a high thermal conductivity polymer composite material assisted by machine learning provided in an embodiment of the present invention comprises the following steps: S1: Building a feature database According to the basic process steps for preparing high thermal conductivity polymer composite materials and the characteristic data related to the target performance, the relevant data on the components, scales, proportions, processes and conditions for preparing high thermal conductivity polymer composite materials are collected to establish a characteristic database; The basic process steps for preparing the high thermal conductivity polymer composite material are: A1: Set the mass ratio range of the main raw materials. The proportion of thermoplastic polymer material ranges from 20-90wt%; carbon fiber 1-30wt%, boron nitride 1-30wt%, additive 1-30wt%; A2: Setting the main preparation process The molding preparation process includes: one of unidirectional freezing, bidirectional freezing, array freezing, and melt blending; S2: Data preprocessing The feature data is converted into binary sub-category features "0" and "1" using one hot encoding to obtain a feature data set; the thermal conductivity of the high thermal conductivity polymer composite material is used as the target variable, and the features in the feature data set are used as independent variables to preprocess the feature data and delete redundant features; the preprocessed feature data set is then divided, with 85% of the feature data set used as a training set and the remaining 15% as a test set; The data encoding in this step is to convert the feature data in step S1 into binary subcategory features "0" and "1", where "0" indicates that it does not belong to the subcategory and "1" indicates that it belongs to the subcategory; S3: Building Machine Learning Models A recursive elimination nested Ridge regression algorithm model based on a machine learning framework is constructed as a machine learning model, and the pre-processed data is used for training to obtain a pre-trained machine learning model as a prediction model; S4: Optimizing the prediction model Five-fold cross validation was used to evaluate the model, optimize parameters, and obtain the optimized Ridge regression prediction model; S5: Prediction model screening The target performance parameters of the high thermal conductivity polymer composite material to be obtained, including thermal conductivity, are input into the prediction model, which performs feature engineering screening on the relevant feature data prepared; specifically, the SHAP nested Ridge regression prediction model based on the machine learning framework is first used for calculation, and then the positive and negative correlation and importance between the independent variable and the target variable are analyzed, and the calculation results of the regression prediction model are explained; S6: Output prediction results According to the prediction results explained by the regression prediction model, the preparation process of high thermal conductivity polymer composite materials is designed and the prediction results are output; S7: Verify the prediction results According to the prediction results, the high thermal conductivity polymer composite material is actually prepared to obtain the composite material, and then the thermal conductivity performance is tested and verified. The test values ​​are compared with the predicted values ​​to verify the accuracy of the model prediction results.

[0028] The S3 constructs a machine learning model, adopts a recursive elimination nested Ridge regression algorithm model as a machine learning model, and uses the pre-processed data for training to obtain a pre-trained machine learning model as a prediction model; the nine features in the database are screened, wherein the nine features are: BN, CF, adhesive, Bidirectional, Array, Blend, unidirectional, Cf content, and Bn content. According to Figure 2 The results show that the coefficient of determination R2 of the model is 0.86 when the number of features is 8. According to the feature screening results, the 8 selected features are BN, CF, Bidirectional, Array, Blend, unidirectional, Cf content and Bn content.

[0029] The target performance-related characteristic data in step S1 include: thermoplastic polymer material components, filler components and content ratios, molding methods, and additives for preparing composite materials; the fillers include carbon fibers and boron nitride.

[0030] The carbon fiber is a polydopamine-coated carbon fiber, which is prepared by the following steps: B1: Ultrasonic washing of carbon fiber in solvent to remove surface impurities, then complete drying in an oven, and then polydopamine coating of the washed carbon fiber; B2: Polydopamine coating process: The pH value of the Tris buffer solution was adjusted to less than 8.5, and then carbon fiber and dopamine were added to the mixed solution in different proportions and reacted at 30-60°C for 12-36 h; B3: The treated carbon fiber is then placed in a beaker and repeatedly washed in deionized water to obtain polydopamine-coated carbon fiber, which is then dried in an oven for later use.

[0031] The boron nitride is hydroxylated boron nitride, which is prepared by the following steps: C1: Boron nitride was hydroxylated as follows: 1.5-3 g of h-BN was added to 100-200 mL of a mixed solution (isopropanol and deionized water) and subjected to tip sonication. C2: The obtained dispersion is then centrifuged at a speed of 500-1000 for 5-20 min, the obtained upper suspension is centrifuged at a speed of 6000-12000 for 5-20 min, and then dried to obtain hydroxylated boron nitride for use.

[0032] The thermoplastic polymer material includes: polyethylene glycol, polyurethane, polycaprolactone, paraffin, thermoplastic polyimide; The fillers are carbon fiber and boron nitride; The auxiliary agents include: a dispersant and a covalent cross-linking agent; The dispersant is a mixture of one or more of deionized water and ethanol; and the covalent cross-linking agent is a mixture of one or more of 1,4-butanediol diglycidyl ether and epichlorohydrin.

[0033] In the step A2, when the melt blending molding method is selected, the filler, the additive and the thermoplastic polymer material are first melt blended at high temperature and the bubbles are removed; then the mixture is placed in a room temperature environment and naturally cooled to solidify the thermoplastic polymer to obtain a high thermal conductivity polymer composite material.

[0034] This embodiment is a basic operation step. In a more specific implementation process, according to the target performance of the high thermal conductivity polymer composite material to be prepared, the specific input data and parameters of the machine learning model are adjusted to obtain the corresponding prediction results, and then the performance compliance experiment is carried out to verify.

[0035] The present invention focuses on collecting data related to target performance based on professional knowledge and establishing a database; preprocessing the data to remove redundant independent variables; dividing the data set, using the Ridge regression algorithm to train and test the divided data set to obtain a prediction model; using five-fold cross validation to evaluate the model, optimizing the parameters to obtain the final prediction model; analyzing the importance of the features; using the final model to optimize the design of the smart polymer composite material and perform experimental verification. The embodiment of the present invention integrates expert knowledge and machine learning, which can accelerate the development of high thermal conductivity smart polymer composite materials, shorten the R&D cycle, and reduce costs.

[0036] Example 2 The method for developing a high thermal conductivity polymer composite material assisted by machine learning provided in the embodiment of the present invention is substantially the same as that in embodiment 1, except that: The step A2, when the one-way freezing molding method is selected for preparation, comprises the following steps: D1: Hydroxylated boron nitride is prepared into a dispersion with a concentration of 10-120 mg / mL, and an acid is used as a catalyst to react with a covalent crosslinking agent for 20-60 min to produce a chemical crosslinking reaction; then the polydopamine-coated carbon fiber is added to the dispersant to prepare a uniformly mixed dispersion; D2: Pour the well-mixed dispersion into a plastic beaker and place it on a cold table immersed in liquid nitrogen for directional freezing.

[0037] D3: Treat with a freeze dryer for 12-96 h and post-treat for 1-6 h to obtain boron nitride / carbon fiber aerogel.

[0038] D4: The thermoplastic polymer material is melted and degassed at high temperature, and vacuum impregnated to fill the voids of the aerogel; it is then placed in a room temperature environment and naturally cooled to solidify the thermoplastic polymer to obtain a high thermal conductivity polymer composite material.

[0039] Example 3 The method for developing a high thermal conductivity polymer composite material assisted by machine learning provided in the embodiment of the present invention is substantially the same as that in embodiment 1, except that: In the step A2, when the bidirectional freezing molding method is selected for preparation: E1: Hydroxylated boron nitride is prepared into a dispersion with a concentration of 10-120 mg / mL, and an acid is used as a catalyst to react with a covalent crosslinking agent for 20-60 min to produce a chemical crosslinking reaction; then, the polydopamine-coated carbon fiber is added to the dispersant to prepare a uniformly mixed dispersion; E2: The uniformly mixed dispersion was poured into a custom-made wedge-shaped mold (the inclination angle of the wedge-shaped mold was 15°) and placed on a cold stage immersed in liquid nitrogen for directional freezing.

[0040] E3: Treat with a freeze dryer for 12-96 h and post-treat for 1-6 h to obtain boron nitride / carbon fiber aerogel.

[0041] E4: The thermoplastic polymer material is melted and degassed at high temperature, and vacuum impregnated to fill the voids of the aerogel; it is then placed in a room temperature environment and naturally cooled to solidify the thermoplastic polymer to obtain a high thermal conductivity polymer composite material.

[0042] Example 4 The method for developing a high thermal conductivity polymer composite material assisted by machine learning provided in the embodiment of the present invention is substantially the same as that in embodiment 1, except that: Step A2, when selecting the array freezing molding method for preparation, comprises the following steps: F1: Hydroxylated boron nitride is prepared into a dispersion with a concentration of 10-120 mg / mL, and an acid is used as a catalyst to react with a covalent crosslinking agent for 20-60 min to produce a chemical crosslinking reaction; then the polydopamine-coated carbon fiber is added to the dispersant to prepare a uniformly mixed dispersion; F2: The well-mixed dispersion was poured into a customized circular copper tube array mold and placed on a cold stage immersed in liquid nitrogen for directional freezing; F3: treated with freeze dryer for 12-96 h and post-treated for 1-6 h to obtain boron nitride / carbon fiber aerogel; F4: The thermoplastic polymer material is melted and degassed at high temperature, and vacuum impregnated to fill the voids of the aerogel; it is then placed in a room temperature environment and naturally cooled to solidify the thermoplastic polymer to obtain a high thermal conductivity polymer composite material.

[0043] Example 5 See also Figure 1-4 As shown in Table 1, in this embodiment, based on the above embodiments, a more specific machine learning-assisted method for developing a high thermal conductivity polymer composite material is provided, the difference being that it further includes the following steps: The target performance-related characteristic data (eigenvalues), predicted output, and measured performance are shown in Table 1.

[0044] The parameter information input in Example 1 is shown in Table 1. The process steps of the high thermal conductivity polymer composite material output by machine learning prediction are: The mass proportion range of the main raw materials is: carbon fiber 9wt%, boron nitride 4.5wt%, thermoplastic polymer material 82wt%; covalent cross-linking agent 4.5wt%, dispersant 6mL.

[0045] The molding preparation process is: array freezing.

[0046] The thermoplastic polymer material is polyethylene glycol; The fillers are carbon fiber and boron nitride; The auxiliary agent includes: deionized water and 1,4-butanediol diglycidyl ether; The high thermal conductivity polymer composite material was prepared and verified according to the following steps: A1: According to the material mass ratio range, a high thermal conductive polymer composite material is prepared, and the ratio range is: carbon fiber 9wt%, boron nitride 4.5wt%, thermoplastic polymer material 82wt%; covalent crosslinking agent 4.5wt%, dispersant 6mL; A2: Preparation of Aerogel A3: Preparation of high thermal conductivity polymer composite materials The thermoplastic polymer is melted at high temperature and bubbles are removed, and the melted polymer is vacuum infused into the interior of the aerogel to fill its internal gaps; it is then placed in a room temperature environment and naturally cooled to solidify the thermoplastic polymer to obtain a high thermal conductivity polymer composite material.

[0047] The thermoplastic polymer material described in step A1 is polyethylene glycol; the molding method is array freezing; the auxiliary agents are: a dispersant (deionized water) and a covalent crosslinking agent (1,4-butanediol diglycidyl ether); Step A2 includes preparing the aerogel: A2-1 Boron nitride is treated with hydroxylation. The specific operation is as follows: 1.5-3 g of h-BN is added to 100-200 mL of a mixed solution (isopropanol and deionized water) for tip ultrasound, and then the obtained dispersion is centrifuged at 500 rpm for 10 min, and the obtained upper suspension is centrifuged at 10000 rpm for 10 min, and then dried to obtain hydroxylated boron nitride; the carbon fiber is ultrasonically washed in a solvent to remove surface impurities, and then completely dried in an oven; the washed carbon fiber is then coated with polydopamine. During the polydopamine coating process, the pH value of the Tris buffer solution is adjusted to less than 8.5, and then the carbon fiber and dopamine are added to the mixed solution in different proportions, and reacted at 60 °C for 24 h. The treated carbon fiber is then placed in a beaker, repeatedly washed in deionized water to obtain polydopamine-coated carbon fiber, which is dried in an oven for later use; A2-2: Hydroxylated boron nitride is prepared into a dispersion with a concentration of 40 mg / mL, and an acid is used as a catalyst to react with a covalent crosslinking agent for 20 min to produce a chemical crosslinking reaction; then, the polydopamine-coated carbon fiber is added to the dispersant to prepare a 120 mg / mL uniformly mixed dispersion; A2-3 Pour the uniformly mixed dispersion into a circular array mold and place it on a cold stage immersed in liquid nitrogen for directional freezing.

[0048] A2-4 was treated with a freeze dryer for 24 h and post-treated for 2.5 h to obtain aerogel.

[0049] The high thermal conductivity polymer composite material is prepared by an array freezing molding method, which specifically includes the following steps: The hydroxylated boron nitride was prepared into a dispersion with a concentration of 40 mg / mL, and an acid was used as a catalyst to react with a covalent crosslinking agent for 20 minutes to produce a chemical crosslinking reaction; then the polydopamine-coated carbon fiber was added to the dispersant to prepare a 120 mg / mL uniformly mixed dispersion; The well-mixed dispersion was poured into a custom-made circular copper tube array mold and placed on a cold stage immersed in liquid nitrogen for directional freezing.

[0050] The aerogel was treated in a freeze dryer for 24 h and post-treated for 2.5 h to obtain boron nitride / carbon fiber aerogel.

[0051] The thermoplastic polymer material is melted and degassed at high temperature, and vacuum impregnated to fill the voids of the aerogel; it is then placed in a room temperature environment and naturally cooled to solidify the thermoplastic polymer to obtain a high thermal conductivity polymer composite material.

[0052] The thermal conductivity of the composite material obtained in this example was tested, and the test value was compared with the predicted value to verify the accuracy of the model prediction. The thermal conductivity of the intelligent carbon fiber high thermal conductivity polymer composite material obtained in Example 1 was tested using the LFA447 laser thermal conductivity instrument produced by the German Netzsch Company. The test results showed that the longitudinal thermal conductivity was 1.8 W • m -1 • K -1 The model predicted value is 1.747 W •m -1 • K -1 , the root mean square error between the experimental value and the predicted value is 0.03732, which proves that the prediction performance of the model is good.

[0053] Therefore, the performance of the high thermal conductivity polymer composite material developed in this embodiment meets the target development requirements.

[0054] Example 6 Referring to Table 1, the method for developing a high thermal conductive polymer composite material assisted by machine learning provided in this embodiment is substantially the same as that in Example 5, except that the characteristic values, test values, and predicted values ​​in Table 1 are different. The parameter information input in Example 2 is shown in Table 1. The high thermal conductivity polymer composite material is prepared according to the following experimental steps: The mass ratio range of the main raw materials output after prediction is: Carbon fiber 11.5wt%, boron nitride 5.7wt%, thermoplastic polymer material 77.1wt%; covalent crosslinking agent 5.7wt%, dispersant 6mL.

[0055] The main preparation process determined is: array freezing, which specifically includes the following steps: Hydroxylated boron nitride was prepared into a dispersion with a concentration of 50 mg / mL, and an acid was used as a catalyst to react with a covalent crosslinking agent for 20 minutes to produce a chemical crosslinking reaction; then, the polydopamine-coated carbon fiber was added to the dispersant to prepare a 150 mg / mL uniformly mixed dispersion; The well-mixed dispersion was poured into a custom-made circular copper tube array mold and placed on a cold stage immersed in liquid nitrogen for directional freezing.

[0056] The aerogel was treated in a freeze dryer for 24 h and post-treated for 2.5 h to obtain boron nitride / carbon fiber aerogel.

[0057] The thermoplastic polymer material is melted and degassed at high temperature, and vacuum impregnated to fill the voids of the aerogel; it is then placed in a room temperature environment and naturally cooled to solidify the thermoplastic polymer to obtain a high thermal conductivity polymer composite material.

[0058] The thermal conductivity of the composite material prepared in this example was tested, and the test value was compared with the predicted value to verify the accuracy of the model prediction. The thermal conductivity of the intelligent carbon fiber high thermal conductivity polymer composite material prepared in Example 1 was tested using the LFA447 laser thermal conductivity instrument produced by the German Netzsch Company. The test results showed that the longitudinal thermal conductivity was 2.1 W • m -1 • K -1 The model predicted value is 2.098 W •m -1 • K -1 , the root mean square error between the experimental value and the predicted value is 0.001246, which proves that the prediction performance of the model is good.

[0059] Table 1 shows the characteristic values ​​of the target materials in specific embodiments 5-6, as well as the experimental values, predicted values ​​and the root mean square error of the two.

[0060] Table 1

[0061] It can be seen that the machine learning-assisted high thermal conductivity polymer composite material development method provided in the above-mentioned embodiments of the present invention, through the coordinated cooperation of steps: S1-S7, can be based on machine learning and prediction of the components and preparation process of the high thermal conductivity polymer composite material, integrate the feature database and machine learning model, accelerate the development of high thermal conductivity intelligent polymer composite materials, shorten the research and development cycle, and do not require high-temperature equipment, thereby reducing preparation costs.

[0062] It should be noted that the above embodiments describe only a part of the embodiments of the present invention, rather than all the embodiments. In other embodiments, within the scope of the present invention, the technical solutions obtained by selecting other characteristic data, components, proportions and process parameters can also achieve the technical effects recorded in the present invention, so they are no longer listed one by one.

[0063] The specific embodiments described above are only an introduction to the optimization method of the present invention, and are not intended to limit the scope of the present invention. It should be noted that, without departing from the spirit of the present invention, those skilled in the art may make appropriate adjustments or modifications to the technical solution of the present invention, which shall fall within the scope of protection determined by the claims of the present invention.

Claims

1. A method for developing a high thermal conductivity polymer composite material assisted by machine learning, characterized in that: It includes the following steps: S1: Building a feature database According to the basic process steps for preparing high thermal conductivity polymer composite materials and the characteristic data related to the target performance, the relevant data on the components, scales, proportions, processes and conditions for preparing high thermal conductivity polymer composite materials are collected to establish a characteristic database; S2: Data preprocessing The feature data is converted into binary sub-category features "0" and "1" using one hot encoding to obtain a feature data set; the thermal conductivity of the high thermal conductivity polymer composite material is used as the target variable, and the features in the feature data set are used as independent variables to pre-process the feature data and delete redundant features; Then divide the preprocessed feature data set into 85% as the training set and the other 15% as the test set; S3: Building Machine Learning Models A recursive elimination nested Ridge regression algorithm model based on a machine learning framework is constructed as a machine learning model, and the pre-processed data is used for training to obtain a pre-trained machine learning model as a prediction model; S4: Optimizing the prediction model Five-fold cross validation was used to evaluate the model, optimize parameters, and obtain the optimized Ridge regression prediction model; S5: Prediction model screening The target performance parameters of the high thermal conductivity polymer composite material to be obtained, including thermal conductivity, are input into the prediction model, which performs feature engineering screening on the relevant feature data prepared; specifically, the SHAP nested Ridge regression prediction model based on the machine learning framework is first used for calculation, and then the positive and negative correlation and importance between the independent variable and the target variable are analyzed, and the calculation results of the regression prediction model are explained; S6: Output prediction results According to the prediction results explained by the regression prediction model, the preparation process of high thermal conductivity polymer composite materials is designed and the prediction results are output; S7: Verify the prediction results According to the prediction results, the high thermal conductivity polymer composite material is actually prepared to obtain the composite material, and then the thermal conductivity performance is tested and verified. The test values ​​are compared with the predicted values ​​to verify the accuracy of the model prediction results.

2. The method for developing a high thermal conductivity polymer composite material assisted by machine learning according to claim 1, characterized in that: The target performance-related characteristic data in step S1 include: thermoplastic polymer material components, filler components and content ratios, molding methods, and additives for preparing composite materials; the fillers include carbon fibers and boron nitride.

3. The method for developing a high thermal conductivity polymer composite material assisted by machine learning according to claim 2, characterized in that: The carbon fiber is a polydopamine-coated carbon fiber, which is prepared by the following steps: B1: Ultrasonic washing of carbon fiber in solvent to remove surface impurities, then complete drying in an oven, and then polydopamine coating of the washed carbon fiber; B2: Polydopamine coating process: The pH value of the Tris buffer solution was adjusted to less than 8.5, and then carbon fiber and dopamine were added to the mixed solution in different proportions and reacted at 30-60°C for 12-36 h; B3: The treated carbon fiber is then placed in a beaker and repeatedly washed in deionized water to obtain polydopamine-coated carbon fiber, which is then dried in an oven for later use.

4. The method for developing a high thermal conductivity polymer composite material assisted by machine learning according to claim 3, characterized in that: The boron nitride is hydroxylated boron nitride, which is prepared by the following steps: C1: Boron nitride is hydroxylated as follows: 1.5-3 g of h-BN is added to 100-200 mL of the mixed solution and subjected to tip sonication. C2: The obtained dispersion is then centrifuged at a speed of 500-1000 for 5-20 min, the obtained upper suspension is centrifuged at a speed of 6000-12000 for 5-20 min, and then dried to obtain hydroxylated boron nitride for use.

5. The method for developing a high thermal conductivity polymer composite material assisted by machine learning according to claim 1, characterized in that: The basic process steps for preparing the high thermal conductivity polymer composite material in step S1 are: A1: Set the mass ratio range of the main raw materials. The proportion of thermoplastic polymer material ranges from 20-90wt%; carbon fiber 1-30wt%, boron nitride 1-30wt%, additive 1-30wt%; A2: Setting the main preparation process The molding preparation processes include: unidirectional freezing, bidirectional freezing, array freezing, and melt blending.

6. The method for developing a high thermal conductivity polymer composite material assisted by machine learning according to claim 5, characterized in that: The thermoplastic polymer material includes: polyethylene glycol, polyurethane, polycaprolactone, paraffin, thermoplastic polyimide; The fillers are carbon fiber and boron nitride; The auxiliary agents include: a dispersant and a covalent cross-linking agent; The dispersant is a mixture of one or more of deionized water and ethanol; and the covalent cross-linking agent is a mixture of one or more of 1,4-butanediol diglycidyl ether and epichlorohydrin.

7. The method for developing a high thermal conductivity polymer composite material assisted by machine learning according to claim 5, characterized in that: In the step A2, when the melt blending molding method is selected, the filler, the additive and the thermoplastic polymer material are first melt blended at high temperature and the bubbles are removed; then the mixture is placed in a room temperature environment and naturally cooled to solidify the thermoplastic polymer to obtain a high thermal conductivity polymer composite material.

8. The method for developing a high thermal conductivity polymer composite material assisted by machine learning according to claim 5, characterized in that: In the step A2, when the one-way freezing molding method is selected for preparation: D1: Hydroxylated boron nitride is prepared into a dispersion with a concentration of 10-120 mg / mL, and an acid is used as a catalyst to react with a covalent crosslinking agent for 20-60 min to produce a chemical crosslinking reaction; then the polydopamine-coated carbon fiber is added to the dispersant to prepare a uniformly mixed dispersion; D2: Pour the well-mixed dispersion into a plastic beaker and place it on a cold table immersed in liquid nitrogen for directional freezing; D3: freeze dry for 12-96 h and post-treat for 1-6 h to obtain boron nitride / carbon fiber aerogel; D4: The thermoplastic polymer material is melted and degassed at high temperature, and vacuum impregnated to fill the voids of the aerogel; it is then placed in a room temperature environment and naturally cooled to solidify the thermoplastic polymer to obtain a high thermal conductivity polymer composite material.

9. The method for developing a high thermal conductivity polymer composite material assisted by machine learning according to claim 5, characterized in that: The step A2, when the bidirectional freezing molding method is selected for preparation, comprises the following steps: E1: Hydroxylated boron nitride is prepared into a dispersion with a concentration of 10-120 mg / mL, and an acid is used as a catalyst to react with a covalent crosslinking agent for 20-60 min to produce a chemical crosslinking reaction; then, the polydopamine-coated carbon fiber is added to the dispersant to prepare a uniformly mixed dispersion; E2: The well-mixed dispersion was poured into a custom-made wedge-shaped mold and placed on a cold stage immersed in liquid nitrogen for directional freezing; E3: treated with freeze dryer for 12-96 h and post-treated for 1-6 h to obtain boron nitride / carbon fiber aerogel; E4: The thermoplastic polymer material is melted and degassed at high temperature, and vacuum impregnated to fill the voids of the aerogel; it is then placed in a room temperature environment and naturally cooled to solidify the thermoplastic polymer to obtain a high thermal conductivity polymer composite material.

10. The method for developing a high thermal conductivity polymer composite material assisted by machine learning according to claim 5, characterized in that: The step A2, when selecting the array freezing molding method for preparation, comprises the following steps: F1: Hydroxylated boron nitride is prepared into a dispersion with a concentration of 10-120 mg / mL, and an acid is used as a catalyst to react with a covalent crosslinking agent for 20-60 min to produce a chemical crosslinking reaction; then the polydopamine-coated carbon fiber is added to the dispersant to prepare a uniformly mixed dispersion; F2: The well-mixed dispersion was poured into a customized circular copper tube array mold and placed on a cold stage immersed in liquid nitrogen for directional freezing; F3: treated with freeze dryer for 12-96 h and post-treated for 1-6 h to obtain boron nitride / carbon fiber aerogel; F4: The thermoplastic polymer material is melted and degassed at high temperature, and vacuum impregnated to fill the voids of the aerogel; it is then placed in a room temperature environment and naturally cooled to solidify the thermoplastic polymer to obtain a high thermal conductivity polymer composite material.

Images