Method for constructing dense membrane structure, training method, prediction method and electronic device

Abstract

This disclosure relates to a method, training method, prediction method, and electronic device for constructing a dense membrane structure. The method includes: constructing a siloxane prepolymer structure model; performing geometric optimization on the siloxane prepolymer structure model and the dimethylsiloxane monomer structure model; randomly placing a first number of geometrically optimized siloxane prepolymer structure models and a second number of geometrically optimized dimethylsiloxane monomer structure models into a pre-sized three-dimensional box for random mixing to obtain an amorphous mixture system; adjusting the atomic coordinates of the amorphous mixture system using a conjugate gradient algorithm to find the stable structure corresponding to the minimum static energy of the amorphous mixture system; and performing temperature equilibrium simulation on the stable structure under a pre-defined ReaxFF force field to form a dimethylsiloxane dense membrane structure. Constructing a dense membrane through molecular dynamics simulation and predicting the degree of crosslinking using artificial intelligence methods can improve research and development efficiency.

Description

Technical Field

[0001] The present disclosure relates to the field of material computing, and in particular, to a method for constructing a dense membrane structure, a training method, a prediction method, and an electronic device. Background Art

[0002] The separation and purification of gases is an indispensable process in industry. Common gas separation methods in the existing chemical industry include adsorption, absorption, membrane separation and condensation gasification. Among them, since different gas molecules have different diffusion rates through the membrane under external pressure, the membrane separation method can be used to separate gases. Compared with other methods, the membrane method has many advantages such as economy, environmental protection, low energy consumption, and easy modification. Therefore, this method currently has broad research prospects in the field of gas separation.

[0003] Membrane separation technology uses the gas partial pressure difference on both sides of the membrane as the driving force, and achieves separation through the different permeation rates of each component in the mixed gas in the gas separation membrane. During the separation process, the gas component with a higher permeation rate (called permeate gas) will quickly migrate through the membrane to the low-pressure side, while the gas component with a lower permeation rate (called retentate gas) will mostly remain on the high-pressure side, thereby achieving the separation of the mixed gas. Polydimethylsiloxane has become one of the commonly used polymers for gas adsorption separation due to its large gas flux, good chemical stability, flexibility and thermal stability. The Si-O-Si chain in polydimethylsiloxane gives it good permeability, allowing the guest molecule to exhibit excellent performance in the structural separation process. In addition, due to the low polarity of the hydrophobic methyl functional group on the side chain, polydimethylsiloxane exhibits good hydrophobicity and organophilicity, which enables the polydimethylsiloxane membrane to preferentially permeate small organic molecules, effectively improving its selectivity in gas adsorption separation.

[0004] Experimental devices still have certain limitations in the structural characterization of dense polymers. The development of computers and mathematical methods has led to the emergence of a series of simulation methods such as molecular dynamics (MD), which has enabled the research level to reach the molecular level. These methods have the advantages of short operation cycle and low cost. MD simulation has been widely used in many fields such as polymer synthesis and new material design, greatly promoting the progress of materials science.

[0005] Unlike porous membranes (such as metal organic frameworks or zeolite materials), PDMS dense membranes do not have a continuous pore structure because the free volume elements in PDMS dense membranes change over time. This change is related to the complexity of the addition reaction mechanism of PDMS and crosslinkers, which makes it difficult to construct a crosslinking model. Therefore, molecular dynamics studies on the addition reaction of PDMS polymers are relatively rare.

[0006] In addition, it was noted that the researchers used a non-crosslinked polydimethylsiloxane model in previous studies. However, in actual production, when polydimethylsiloxane membranes are used to recover biofuels from dilute organic aqueous solutions, the polydimethylsiloxane chains are chemically crosslinked to improve mechanical properties and resistance to swelling. The degree of crosslinking of the silicone network may be a very important parameter that determines the permeability characteristics of nanofiltration membranes. However, there is still a lack of systematic research on the effect of polydimethylsiloxane crosslinking on the gas adsorption separation performance of silicon-based nanofiltration membranes, mainly because the membranes studied are commercial and the exact nature of their selective layers has not been fully revealed. Nevertheless, the degree of crosslinking of polydimethylsiloxane may play an important role in gas adsorption separation, but further research is needed to systematically evaluate its specific impact. At present, the degree of crosslinking is obtained by performing relevant statistics on the constructed dense membrane. It takes a lot of computing power and time to construct a dense membrane. If the degree of crosslinking of the dense membrane can be predicted in advance based on relevant conditions, it will save time and computing power, guide subsequent research, and improve R&D efficiency. Summary of the invention

[0007] The purpose of the present disclosure is to provide a method, training method, prediction method and electronic device for constructing a dense membrane structure, aiming to construct a polydimethylsiloxane dense membrane with real physical property characteristics based on molecular dynamics simulation, and to train a prediction model that can use molecular dynamics simulation calculation parameters to predict the cross-linking degree of the polydimethylsiloxane dense membrane, so as to predict its cross-linking degree without constructing a dense membrane.

[0008] In order to achieve the above object, the present disclosure provides a method for constructing a dense polydimethylsiloxane film structure in a first aspect, the method comprising:

[0009] Constructing the structural model of siloxane prepolymer and dimethylsiloxane monomer;

[0010] Performing geometric optimization on the siloxane prepolymer structure model and the dimethylsiloxane monomer structure model;

[0011] A first number of geometrically optimized siloxane prepolymer structure models and a second number of geometrically optimized dimethylsiloxane monomer structure models are randomly placed in a three-dimensional box of a preset size for random mixing to obtain an amorphous mixed system;

[0012] Using a conjugate gradient algorithm to adjust the atomic coordinates of the amorphous mixed system, and find a stable structure corresponding to the minimum static energy of the amorphous mixed system;

[0013] For the stable structure, under a preset ReaxFF force field, based on a preset Verlet-Velocity algorithm for integrating the equation of motion and a preset Nose-Hoover algorithm for adjusting the temperature, a temperature equilibrium simulation is performed at any temperature between 500K and 800K, and then a temperature equilibrium simulation is performed at any temperature between 298K and 330K, so that the siloxane prepolymer structure model in the stable structure is connected to the silicon atoms and oxygen atoms of the dimethylsiloxane monomer structure model through its own oxygen active sites and silicon active sites, respectively, to form a dimethylsiloxane dense film structure.

[0014] Optionally, the temperature balance simulation is performed at any temperature between 500K and 800K based on a preset Verlet-Velocity algorithm for integrating the equation of motion and a preset Nose-Hoover algorithm for adjusting the temperature, and then the temperature balance simulation is performed at any temperature between 298K and 330K, including:

[0015] Based on the Verlet-Velocity algorithm with a preset time step of 0.1 fs for integration of the equations of motion and the Nose-Hoover algorithm with a preset standard relaxation time of 10 fs for adjusting the temperature, a temperature equilibrium simulation was performed at 500 K for 50 ps and then at 300 K for 100 ps.

[0016] Optionally, the construction of the siloxane prepolymer structure model and the dimethylsiloxane monomer structure model comprises:

[0017] Constructing a dimethylsiloxane monomer structure model and a tetraethyl silicate structure model, and connecting the oxygen atom of one dimethylsiloxane monomer structure model with the silicon atom of another dimethylsiloxane monomer structure model to obtain a dimethylsiloxane dimer model;

[0018] Two ethoxy groups of the four ethoxy groups of the tetraethyl silicate structural model are removed, and the ethyl groups of the other two ethoxy groups are removed to obtain an intermediate molecular model, and the silicon atom of the intermediate molecular model is connected to the oxygen atoms containing unsaturated bonds of two dimethylsiloxane dimer models, and the two oxygen atoms of the intermediate molecular model are respectively connected to the silicon atom containing unsaturated bonds of a dimethylsiloxane dimer model to obtain a siloxane prepolymer structural model.

[0019] Optionally, geometric optimization is performed on the siloxane prepolymer structure model and the dimethylsiloxane monomer structure model, comprising:

[0020] In Materials Studio software, the Geometry optimization task in the Forcite module was used, the Smart algorithm was adopted, the Energy parameter was set to 0.001 kcal / mol, the Force parameter was set to 0.5 kcal / mol / Å, the Displacement parameter was set to 0.015 Å, the maximum number of iterations was set to 500 steps, the force field used was COMPASSIII force field, the charge parameter was set to Forcefield assigned, the charge group was set to automatic force field assignment, the electrostatic term was set to Ewald summation method, the accuracy parameter was set to 0.001 kcal / mol, the van der Waals term was set to Atom-based summation method, the cutoff method used was Cubic spline, the cutoff radius was set to 12.5 Å, the bond width was set to 1Å, and Long range correction was used for geometry optimization during the calculation process.

[0021] Optionally, the second number is any multiple between 8 and 40 times the first number.

[0022] Optionally, the first number is 30 and the second number is 274.

[0023] Optionally, the method further comprises:

[0024] Select the NPT ensemble, set the pressure to 1000 bar, the temperature to 300 K, and the simulation time to 300 ps, ​​and perform a volume compression simulation calculation based on molecular dynamics simulation on the dimethylsiloxane dense membrane structure;

[0025] Select the NVT ensemble, set the temperature to 800K, and the simulation time to 100ps, and perform a heating relaxation simulation calculation based on molecular dynamics simulation on the dimethylsiloxane dense membrane structure that has undergone volume compression;

[0026] Select the NVT ensemble, set the temperature to 300K, and the simulation time to 100ps, and perform a cooling simulation calculation based on molecular dynamics simulation on the dimethylsiloxane dense membrane structure after heating and relaxation;

[0027] Select the NPT ensemble, set the pressure to 100 bar, the temperature to 300 K, and the simulation time to 300 ps, ​​and perform a recompression simulation calculation based on the molecular dynamics simulation on the dimethylsiloxane dense membrane structure that has undergone the cooling simulation calculation;

[0028] When the number of iterative calculations is less than the threshold and the density difference of the dimethylsiloxane dense film obtained by two adjacent recompression simulations is greater than 0.001 g / cm 3, returning to the step of selecting the NPT ensemble, setting the pressure to 1000 bar, the temperature to 300 K, the simulation time to 300 ps, ​​and performing a volume compression simulation calculation based on molecular dynamics simulation on the dimethylsiloxane dense membrane structure;

[0029] When the number of iterative calculations is greater than or equal to the threshold or the density difference of the dimethylsiloxane dense film obtained by two adjacent recompression simulations is less than or equal to 0.001 g / cm 3 When the NPT ensemble is selected, the pressure is set to 1 bar, the temperature is set to 300 K, and the simulation time is set to 10 ns. The molecular dynamics simulation calculation is performed on the dimethylsiloxane dense membrane structure obtained by the final compression simulation calculation to obtain a stable dimethylsiloxane dense membrane structure.

[0030] In a second aspect of the present disclosure, a random forest model training method is provided, the method comprising:

[0031] Setting a plurality of different groups of sample data, wherein the sample data includes the number of siloxane prepolymer structure models, the number of dimethylsiloxane monomer structure models, and at least part of the temperature and pressure parameters in the method according to the first aspect;

[0032] Apply each set of sample data to the method described in the first aspect and execute the method, and after the method is executed, obtain multiple sets of lengths, widths and heights of the three-dimensional box;

[0033] The degree of crosslinking is calculated based on the length, width and height using the following formula:

[0034] ;

[0035] Wherein, CKI represents the degree of cross-linking, X, Y and Z are the length, width and height of the three-dimensional box, respectively, and n is the number of siloxane prepolymer structure models;

[0036] The cross-linking degree and corresponding sample data form a training data set, and a random forest model is trained based on the training data set to obtain a trained random forest model for predicting the cross-linking degree.

[0037] In a third aspect of the present disclosure, a method for predicting a degree of crosslinking based on artificial intelligence is provided, the method comprising:

[0038] Obtaining temperature, pressure, number of siloxane prepolymer structure models, and number of dimethylsiloxane monomer structure models;

[0039] The obtained temperature, pressure, number of siloxane prepolymer structure models and number of dimethylsiloxane monomer structure models are input into the random forest model trained by the method described in the second aspect to obtain the predicted output of the crosslinking degree of the dimethylsiloxane dense film structure.

[0040] According to a fourth aspect of the present disclosure, there is provided an electronic device, comprising:

[0041] a memory having a computer program stored thereon;

[0042] A processor is used to execute the computer program in the memory to implement the steps of the method in any one of the first aspect, the second aspect or the third aspect.

[0043] Through the above technical solution, after constructing the siloxane prepolymer structure model and the dimethylsiloxane monomer structure model respectively, the geometry optimization is first performed, and then the two models are randomly mixed to obtain an amorphous mixed system. Then, the atomic coordinates of the amorphous mixed system are adjusted using the conjugate gradient algorithm to find its stable structure with the minimum static energy. Finally, the ReaxFF force field, Verlet-Velocity algorithm and Nose-Hoover algorithm are used to perform molecular dynamics simulation calculations to obtain a dimethylsiloxane dense membrane structure. The dimethylsiloxane dense membrane structure obtained through this process is relatively stable and has real physical property characterization, which can be used for subsequent other performance analysis calculations. In addition, the relevant parameters for constructing dimethylsiloxane dense membranes are used as sample data, and the length, width and height of the three-dimensional box obtained after the calculation are completed are used to calculate the crosslinking degree. Based on the crosslinking degree and the corresponding sample data to form a training data set, a random forest model is trained to obtain a prediction model for predicting the crosslinking degree, which is convenient for predicting the crosslinking degree in advance without building a model, saving time and computing power, and improving R&D efficiency.

[0044] Other features and advantages of the present disclosure will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The accompanying drawings are used to provide a further understanding of the present disclosure and constitute a part of the specification. Together with the following specific embodiments, they are used to explain the present disclosure but do not constitute a limitation of the present disclosure. In the accompanying drawings:

[0046] Figure 1 The present invention is a flow chart showing a method for constructing a dense polydimethylsiloxane film structure according to an exemplary embodiment.

[0047] Figure 2 is a diagram showing a structural model of a dimethylsiloxane monomer according to an exemplary embodiment.

[0048] Figure 3 is a diagram of a structural model of tetraethyl silicate according to an exemplary embodiment.

[0049] Figure 4 is a diagram of a siloxane prepolymer structure model according to an exemplary embodiment.

[0050] Figure 5 is a diagram of a dimethylsiloxane monomer structure model after geometric optimization according to an exemplary embodiment.

[0051] Figure 6 is a diagram of a geometrically optimized siloxane prepolymer structure model according to an exemplary embodiment.

[0052] Figure 7 is a diagram of an amorphous mixed system model according to an exemplary embodiment.

[0053] Figure 8 is a diagram showing a model of a dimethylsiloxane dense film structure according to an exemplary embodiment.

[0054] Fig. 9 is another flow chart of a method for constructing a dense polydimethylsiloxane film structure according to an exemplary embodiment.

[0055] Fig.10 The figure is a flowchart of a random forest model training method according to an exemplary embodiment.

[0056] Fig.11 The present invention is a flowchart of a method for predicting cross-linking degree based on artificial intelligence according to an exemplary embodiment.

[0057] Fig.12 It is a block diagram of an electronic device according to an exemplary embodiment. DETAILED DESCRIPTION

[0058] The specific implementation of the present disclosure is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described herein is only used to illustrate and explain the present disclosure, and is not used to limit the present disclosure.

[0059] Molecular dynamics (MD) is an interdisciplinary technology that integrates knowledge of physics, mathematics, and chemistry. It simulates the dynamic behavior of molecular systems by applying the principles of Newtonian mechanics. This method aims to approximate the configuration integral of the system by extracting samples of the molecular system in different states, and then evaluate the thermodynamic properties and other macroscopic characteristics of the system based on these integral results. It allows researchers to study the behavior of complex systems at the atomic and molecular scales, thereby enhancing the understanding of natural phenomena. It can be widely used in research in chemistry, physics, materials science and other fields.

[0060] In the present disclosure, the construction of the dense film structure of polydimethylsiloxane uses two basic molecular units, namely dimethylsiloxane monomer and tetraethyl silicate monomer, wherein the molecular formula of dimethylsiloxane is -(Si(CH3)2-O-) n-, whose structural formula is as follows:

[0061]

[0062] The molecular formula of tetraethyl silicate monomer is Si(OEt)4(C8H 20 O4Si), its structural formula is as follows:

[0063]

[0064] Figure 1 is a flow chart of a method for constructing a dense polydimethylsiloxane film structure according to an exemplary embodiment. Figure 1 As shown, the method includes:

[0065] S101, constructing the siloxane prepolymer structure model and the dimethylsiloxane monomer structure model.

[0066] S102, geometrically optimizing the siloxane prepolymer structure model and the dimethylsiloxane monomer structure model.

[0067] S103, selecting a first number of geometrically optimized siloxane prepolymer structure models and a second number of geometrically optimized dimethylsiloxane monomer structure models, randomly placing them in a three-dimensional box of a preset size for random mixing to obtain an amorphous mixed system.

[0068] S104, using a conjugate gradient algorithm to adjust the atomic coordinates of the amorphous mixed system, and to find a stable structure corresponding to the minimum static energy of the amorphous mixed system.

[0069] S105. For the stable structure, under a preset ReaxFF force field, based on a preset Verlet-Velocity algorithm for integrating the equation of motion and a preset Nose-Hoover algorithm for adjusting the temperature, a temperature equilibrium simulation is performed at any temperature between 500K and 800K, and then a temperature equilibrium simulation is performed at any temperature between 298K and 330K, so that the siloxane prepolymer structure model in the stable structure is connected to the silicon atoms and oxygen atoms of the dimethylsiloxane monomer structure model through its own oxygen active sites and silicon active sites, respectively, to form a dimethylsiloxane dense film structure.

[0070] In step S101, chemical modeling software such as Material Studio can be used to construct a siloxane prepolymer structure model and a dimethylsiloxane monomer structure model. The siloxane prepolymer structure model is initially constructed using a tetraethyl silicate structure model and a dimethylsiloxane monomer structure model. The siloxane prepolymer structure model serves as an initial core structure and is used to connect more dimethylsiloxane monomer structure models to form a dimethylsiloxane dense film structure.

[0071] Optionally, in step S101, constructing a siloxane prepolymer structure model and a dimethylsiloxane monomer structure model includes:

[0072] S1011, constructing a dimethylsiloxane monomer structure model and a tetraethyl silicate structure model, and connecting the oxygen atom of one dimethylsiloxane monomer structure model with the silicon atom of another dimethylsiloxane monomer structure model to obtain a dimethylsiloxane dimer model.

[0073] S1012, removing two ethoxy groups from the four ethoxy groups of the tetraethyl silicate structural model, and removing the ethyl groups from the other two ethoxy groups to obtain an intermediate molecular model, connecting the silicon atom of the intermediate molecular model to the oxygen atoms containing unsaturated bonds of two dimethylsiloxane dimer models, and connecting the two oxygen atoms of the intermediate molecular model to the silicon atom containing unsaturated bonds of one dimethylsiloxane dimer model, respectively, to obtain a siloxane prepolymer structural model.

[0074] In step S1011, a dimethylsiloxane monomer structure model and a tetraethyl silicate structure model can be constructed in a chemical modeling software such as Material Studio. For example, the two structure models can be constructed respectively according to the atomic spatial arrangement structure in the molecule. The constructed dimethylsiloxane monomer structure model can be found in Figure 2 The structure model of tetraethyl silicate can be found in Figure 3 .

[0075] In addition, in a possible embodiment, in order to construct a single siloxane prepolymer structural model, 8 dimethylsiloxane monomer structural models can be constructed first, and for every 2 dimethylsiloxane monomer structural models, the oxygen atom of one dimethylsiloxane monomer structural model is connected to the silicon atom of another dimethylsiloxane monomer structural model to obtain a total of 4 dimethylsiloxane dimer models.

[0076] As an important cross-linking agent, the terminal ester group of tetraethyl silicate reacts with the terminal hydroxyl group of the polydimethylsiloxane chain to form a three-dimensional network structure cross-linked molecule, which gives polydimethylsiloxane a wide range of application prospects.

[0077] After step S1011 is executed, step S1012 may be executed. In step S1012, the tetraethyl silicate structural model includes four ethoxy groups, two of which are directly removed, and the ethyl groups in the other two ethoxy groups are removed respectively to obtain an intermediate molecular model, which is composed of a silicon atom connected to two oxygen atoms.

[0078] Then, the oxygen atoms containing unsaturated bonds of the two dimethylsiloxane dimer models are connected to the silicon atoms of the intermediate molecule model, and the silicon atoms containing unsaturated bonds of the other two dimethylsiloxane dimer models are respectively connected to the two oxygen atoms of the intermediate molecule model to obtain a siloxane prepolymer structure model, the structural formula of which is as follows:

[0079]

[0080] The constructed siloxane prepolymer structure model can be found in Figure 4 .

[0081] After step S101 is completed, step S102 may be executed. In step S102, material calculation software such as Material Studio may be used to perform geometric optimization on the constructed siloxane prepolymer structure model to obtain a stable configuration of the siloxane prepolymer structure model.

[0082] Optionally, in step S102, geometric optimization is performed on the siloxane prepolymer structure model and the dimethylsiloxane monomer structure model, including:

[0083] In Materials Studio software, the Geometry optimization task in the Forcite module was used, the Smart algorithm was adopted, the Energy parameter was set to 0.001 kcal / mol, the Force parameter was set to 0.5 kcal / mol / Å, the Displacement parameter was set to 0.015 Å, the maximum number of iterations was set to 500 steps, the force field used was COMPASSIII force field, the charge parameter was set to Forcefield assigned, the charge group was set to automatic force field assignment, the electrostatic term was set to Ewald summation method, the accuracy parameter was set to 0.001 kcal / mol, the van der Waals term was set to Atom-based summation method, the cutoff method used was Cubic spline, the cutoff radius was set to 12.5 Å, the bond width was set to 1Å, and Long range correction was used for geometry optimization during the calculation process.

[0084] Specifically, you can use the relevant calculation modules in Materials Studio software to perform geometry optimization calculations based on molecular dynamics simulations. You can select the Geometry Optimization task in the Forcite module, and select the relevant parameters as described above. This provides a balanced accuracy and efficiency for geometry optimization, which is not only suitable for processing complex macromolecules and periodic systems, but also can optimize the use of computing resources while maintaining high accuracy. The structure of the dimethylsiloxane monomer after geometry optimization can be obtained by referring to Figure 5 The geometrically optimized siloxane prepolymer structure model can be found in Figure 6 .

[0085] Of course, in other implementations, other methods may be selected for geometry optimization, for example, DFTB+ and LAMMPS may be used to write geometry optimization scripts.

[0086] In step S103, the second number is any multiple between 8 and 40 times of the first number, for example, 8, 9 or 40 times. In a possible implementation, the first number is 30 and the second number is 274. The preset size of the three-dimensional box may be 50Å×50Å×50Å. Figure 7 The cube formed by the lines shown in the figure is a three-dimensional box, which is used to represent the core component of the crystal structure - the unit cell. It is the basic building block that reveals the orderly and periodic arrangement rules inside the crystal. In the three-dimensional space coordinate system, the unit cell is precisely defined by three mutually perpendicular basis vectors, which together form a framework structure that may be a cube or other polyhedron in geometric form, forming a cubic framework structure here. By arranging atoms or molecules in such a three-dimensional box, a small part of the atomic distribution in the crystal, namely the first Bravais zone, can be accurately simulated; and this small part is continuously replicated and extended according to the periodic rules of the lattice, so that the unbounded and regularly arranged structural characteristics of the entire crystal can be fully reproduced.

[0087] For example, if the geometry optimization is performed using Materials Studio software in step S102, the atomic structure information file of the geometry-optimized siloxane prepolymer structure model and the dimethylsiloxane monomer structure model in pdb format can be exported after the geometry optimization is completed. Then the atomic structure information file is imported into Packmol software, and 30 geometry-optimized siloxane prepolymer structure models and 274 geometry-optimized dimethylsiloxane monomer structure models can be randomly placed in a three-dimensional box with a size of 50Å×50Å×50Å, as reference. Figure 7 , to simulate an amorphous mixed system under actual conditions. This method allows dimethylsiloxane monomers and siloxane prepolymers to be randomly distributed in the same simulation environment, thereby simulating their interactions and structure formation processes. By randomly mixing and placing, the initial structure of the material can be accurately established, providing appropriate starting conditions for subsequent molecular dynamics simulations to deeply explore the changes in its structure and properties.

[0088] After step S103 is completed, the process may proceed to step S104. In step S104, a conjugate gradient (cg) algorithm may be used to perform static energy minimization processing on the amorphous mixed system.

[0089] For example, LAMMPS software can be used for static minimization. If the amorphous mixed system is obtained by random mixing using Packmol software in step S103, the molecular structure information file of the amorphous mixed system can be exported. Since the molecular structure information file exported by Packmol software is in pdb format, it is not compatible with LAMMPS software, so it is first converted into a format.

[0090] Optionally, the molecular structure information file in pdb format can be converted into lmp format file using Atomsk software, and then the lmp format file can be converted into data format file readable by LAMMPS software using OVITO software. In this way, a seamless transition from construction to simulation of the initial model is ensured, so that the molecular dynamics simulation can proceed smoothly, providing a reliable basis for subsequent simulation and analysis.

[0091] Using the above example, to implement calculations using LAMMPS software, first write an adapted in format file according to the simulation format requirements and integrate it into the script to form an automatically executable file. This method not only simplifies the simulation process and improves efficiency, but also ensures that each step runs automatically according to the preset parameters and conditions, thereby ensuring the accuracy and stability of the simulation.

[0092] The calculation script is then run using LAMMPS software to read the data format file, load the amorphous mixed system, and then fine-tune the atomic coordinates of the loaded amorphous mixed system based on the conjugate gradient algorithm to find the most stable structure of the system. The conjugate gradient algorithm is an efficient global optimization technique suitable for unconstrained optimization problems, known for its iterative simplicity and low storage requirements. This optimization process is crucial to ensuring the convergence and accuracy of subsequent molecular dynamics simulations.

[0093] After step S104 is completed, step S105 can be executed. In order to better understand and simulate the cross-linking process of polydimethylsiloxane, a reactive force field (ReaxFF) can be used. As a new type of force field, ReaxFF determines the connectivity between atoms by calculating the bond order between any two atoms at the current moment. In the MD-ReaxFF simulation, the chemical bonds between atoms can be broken and generated freely, and as the chemical bonds are broken and generated, the connectivity between atoms is also updated. This feature enables ReaxFF to dynamically capture the breaking of chemical bonds, the formation of bonds, and specific chemical reactions, thereby realizing the simulation of chemical reaction systems. This force field is suitable for systems containing the same type of elements and can be used for the study of silicone materials. The application of this force field to the simulation of dimethylsiloxane dense film is reasonable and applicable, and can accurately describe the chemical and physical properties of the material, ensuring the reliability and accuracy of the simulation results.

[0094] This characteristic of ReaxFF solves the problem that traditional molecular dynamics (MD) simulations cannot simulate the dynamic chemical reactions of the system. At the same time, ReaxFF simulations can simulate larger-scale systems with smaller computing resources. Its force field parameters are completely derived from the fitting of the quantum chemical calculation results of the force field target system. It can quickly and accurately observe the time evolution trends of reactants, intermediates and products in the chemical process, providing an effective way to study the physicochemical properties of larger reaction systems. After years of development, the ReaxFF reaction force field has been widely used in many research fields such as energetic materials, catalysis, combustion and polymers. Therefore, using ReaxFF to simulate the cross-linking process of PDMS can better understand the formation of its three-dimensional network structure and its application potential in gas adsorption and separation.

[0095] Optionally, in step S105, based on a preset Verlet-Velocity algorithm for integrating the equation of motion and a preset Nose-Hoover algorithm for adjusting the temperature, a temperature balance simulation is performed at any temperature between 500K and 800K, and then a temperature balance simulation is performed at any temperature between 298K and 330K, including:

[0096] Based on the Verlet-Velocity algorithm with a preset time step of 0.1 fs for integration of the equations of motion and the Nose-Hoover algorithm with a preset standard relaxation time of 10 fs for adjusting the temperature, a temperature equilibrium simulation was performed at 500 K for 50 ps and then at 300 K for 100 ps.

[0097] Specifically, the NVT ensemble was adopted, and the Verlet-Velocity algorithm was used to integrate the equations of motion with a time step of 0.1 fs. The Nose-Hoover algorithm was used to adjust the temperature, and the standard relaxation time was 10 fs. First, a 50ps temperature equilibrium was performed at 500K with a step size of 0.1fs; then, a 100ps equilibrium simulation was continued at 300K to further stabilize the system. During the simulation process, the system data file in data format and the trajectory file in XYZ format can be output through the LAMMPS command to record and analyze the dynamic evolution of the system. The automated script was executed, and the new siloxane prepolymer structure was successfully constructed by repeatedly repeating the molecular dynamics simulation at different temperatures under NVT. These prepolymers are connected to the terminal silicon atoms and oxygen atoms in the dimethylsiloxane monomer structure through two oxygen active sites and two silicon active sites to form a stable three-dimensional network structure. In one embodiment, the constructed dimethylsiloxane dense membrane structure is as follows Figure 8 shown.

[0098] Through the above technical scheme, after constructing the siloxane prepolymer structure model and the dimethylsiloxane monomer structure model respectively, the geometric optimization is first performed, and then the two models are randomly mixed to obtain an amorphous mixed system. Then, the conjugate gradient algorithm is used to adjust the atomic coordinates of the amorphous mixed system to find its stable structure with minimum static energy. Finally, the ReaxFF force field, Verlet-Velocity algorithm and Nose-Hoover algorithm are used to perform molecular dynamics simulation calculations to obtain the dimethylsiloxane dense membrane structure. The dimethylsiloxane dense membrane structure obtained through this process is relatively stable and has real physical property characterization, which can be used for subsequent other performance analysis calculations.

[0099] In addition, through the above technical solution, a method for successfully constructing a dense polydimethylsiloxane film was used using the ReaxFF reaction force field, and the precise construction and evaluation of polymer materials was achieved through molecular dynamics simulation. The application of the ReaxFF force field can not only simulate the complex interactions between dimethylsiloxane molecules, but also effectively capture its structure and mechanical properties, providing an in-depth understanding and new exploration paths for material design and engineering applications. Through simulation, the structural evolution and dynamic behavior of polydimethylsiloxane molecules were deeply explored, revealing the cross-linking process of siloxane prepolymers and dimethylsiloxane monomers and the final three-dimensional network structure. This method provides an in-depth microscopic perspective for understanding the internal mechanism of the material, making up for the limitations of experimental methods in obtaining detailed information.

[0100] The method disclosed in this disclosure is expected to be further expanded to more complex multi-component material systems to explore its application potential in the fields of nanomaterials and biomedicine. The innovation of this method is that it can accurately describe the structural evolution and dynamic behavior of materials, providing a reliable theoretical basis and experimental guidance for the development of new polydimethylsiloxane materials and their applications. In the future, with the enhancement of computing power and the advancement of simulation technology, these simulation technologies will not only help to accurately predict the performance of materials under different conditions, but also provide key guidance for the design and development of new materials. Optimizing the structure and performance of materials through simulation will help accelerate the experimental verification and market application of new materials.

[0101] In a possible embodiment, after executing step S105 in the above-mentioned method for constructing a dense film structure of polydimethylsiloxane, the structural model and its motion trajectory obtained by molecular dynamics calculation can be analyzed in detail by visualization software. The main criterion for evaluating the bonding of polydimethylsiloxane is to observe the connection of the backbone Si-O-Si bond in polydimethylsiloxane, and the trajectory file generated by simulation using Materials Studio or VMD software analysis, such as a coordinate file in XYZ format, can intuitively display the dynamic path of the reaction, especially the cross-linking process between the siloxane prepolymer and the dimethylsiloxane monomer. Carbon atoms and hydrogen atoms can be selectively hidden to more clearly observe the structure of the backbone Si-O-Si bond, and when the connection of the Si-O-Si bond meets expectations, such as the number of connections meets expectations, it can be determined that the expected dense film structure of dimethylsiloxane is obtained. Of course, the connection of the backbone Si-O-Si bond can also be evaluated by the degree of cross-linking.

[0102] Fig. 9 is another flow chart of a method for constructing a dense polydimethylsiloxane film structure according to an exemplary embodiment. Fig. 9 As shown, the method includes:

[0103] S901, constructing the siloxane prepolymer structure model and the dimethylsiloxane monomer structure model.

[0104] S902, geometrically optimizing the siloxane prepolymer structure model and the dimethylsiloxane monomer structure model.

[0105] S903, selecting a first number of geometrically optimized siloxane prepolymer structure models and a second number of geometrically optimized dimethylsiloxane monomer structure models, randomly placing them in a three-dimensional box of a preset size for random mixing to obtain an amorphous mixed system.

[0106] S904, using a conjugate gradient algorithm to adjust the atomic coordinates of the amorphous mixed system, and to find a stable structure corresponding to the minimum static energy of the amorphous mixed system.

[0107] S905. For the stable structure, under a preset ReaxFF force field, based on a preset Verlet-Velocity algorithm for integrating the equation of motion and a preset Nose-Hoover algorithm for adjusting the temperature, a temperature equilibrium simulation is performed at any temperature between 500K and 800K, and then a temperature equilibrium simulation is performed at any temperature between 298K and 330K, so that the siloxane prepolymer structure model in the stable structure is connected to the silicon atoms and oxygen atoms of the dimethylsiloxane monomer structure model through its own oxygen active sites and silicon active sites, respectively, to form a dimethylsiloxane dense film structure.

[0108] S906, select the NPT ensemble, set the pressure to 1000 bar, the temperature to 300 K, the simulation time to 300 ps, ​​and perform a volume compression simulation calculation based on molecular dynamics simulation on the dimethylsiloxane dense membrane structure.

[0109] S907, select the NVT ensemble, set the temperature to 800K, set the simulation time to 100ps, and perform a heating relaxation simulation calculation based on molecular dynamics simulation on the dimethylsiloxane dense membrane structure that has undergone volume compression.

[0110] S908, select the NVT ensemble, set the temperature to 300K, set the simulation time to 100ps, and perform a cooling simulation calculation based on molecular dynamics simulation on the dimethylsiloxane dense membrane structure that has been relaxed by heating.

[0111] S909, select the NPT ensemble, set the pressure to 100 bar, the temperature to 300K, the simulation time to 300 ps, ​​and perform a recompression simulation calculation based on molecular dynamics simulation on the dimethylsiloxane dense membrane structure that has undergone cooling simulation calculation.

[0112] S910: When the number of iterative calculations is less than the threshold and the density difference of the dimethylsiloxane dense film obtained by two adjacent recompression simulations is greater than 0.001 g / cm 3 When the NPT ensemble is selected, the pressure is set to 1000 bar, the temperature is set to 300 K, the simulation time is set to 300 ps, ​​and the volume compression simulation calculation based on molecular dynamics simulation is performed on the dimethylsiloxane dense membrane structure.

[0113] S911: When the number of iterative calculations is greater than or equal to the threshold number or the density difference of the dimethylsiloxane dense film obtained by two adjacent recompression simulation calculations is less than or equal to 0.001 g / cm 3When the NPT ensemble is selected, the pressure is set to 1 bar, the temperature is set to 300 K, and the simulation time is set to 10 ns. The molecular dynamics simulation calculation is performed on the dimethylsiloxane dense membrane structure obtained by the final compression simulation calculation to obtain a stable dimethylsiloxane dense membrane structure.

[0114] Specifically, steps S901 to S905 may refer to steps S101 to S105.

[0115] After step S905 is executed, step S906 can be executed. In step S906, simulation is performed under the NPT (constant pressure and constant temperature) ensemble, with the pressure set to 1000 bar, the temperature set to 300 K, and the simulation time set to 300 ps. This process is intended to compress the model to the expected volume range through high pressure conditions.

[0116] After step S906 is completed, step S907 can be executed. In step S907, under the NVT (constant volume and constant temperature) ensemble, the temperature is set to 800 K and the simulation time is set to 100 ps. This step allows the molecules to move fully at high temperatures, thereby improving the structural relaxation state of the model, promoting the rearrangement of molecules and reaching a lower energy state.

[0117] After step S907 is completed, step S908 can be executed. In step S908, simulation is performed under the NVT ensemble, the temperature is set to 300 K, and the simulation time is 100 ps. This step gradually cools the model to room temperature to simulate the cooling process in the actual environment, and fixes the structure after high temperature treatment, which helps to solidify the structure of the model under room temperature conditions.

[0118] After step S908 is completed, step S909 can be executed. In step S909, simulation is performed under the NPT ensemble, and the pressure is set to 100 bar, the temperature is set to 300 K, and the simulation time is set to 300 ps. This is to further compress the model under high pressure and simulate the structural changes in the actual high-pressure environment. This step is used to verify and adjust the volume and structural stability of the model under different pressure conditions.

[0119] Steps S906 to S908 may be iteratively performed multiple times to further optimize the structural stability of the model and reduce the residual stress in the structure through multiple heating and cooling processes.

[0120] After step S909 is executed, step S910 may be executed. In step S910, if the number of iterative calculations returned is less than a threshold value such as 30 times, or the density difference of the dimethylsiloxane dense film obtained by the compression simulation calculation in two adjacent iterative processes is greater than 0.001 g / cm 3The obtained model does not meet the requirements and needs to be further iterated. At this time, the process can return to step S906 to perform iterative calculation.

[0121] After step S909 is executed, step S911 may be executed. In step S911, if the number of iterative calculations returned is greater than or equal to the number threshold, such as 30 times, or the density difference of the dimethylsiloxane dense film obtained by the compression simulation calculation in two adjacent iterative processes is less than or equal to 0.001 g / cm 3 , a long-term simulation can be performed under the NPT ensemble, with the pressure set to 1 bar, the temperature to 300 K, and the simulation time to 10 ns to ensure that the model is stable under standard environmental conditions and eventually forms a two-dimensional membrane structure.

[0122] In the related art, the model in the simulation adopts three-dimensional periodic boundary conditions, which means that each boundary of the model is periodic, that is, the simulated area is considered to be repeated infinitely. One limitation of this setting is that there is no actual interface in the simulation because the boundaries of the model are constantly repeated, which shows that the model structure is not suitable for simulating the surface adsorption phenomenon of actual membranes.

[0123] Through the above technical solutions, better structural stability is achieved in the simulation, so that under periodic boundary conditions, the structure of the model can be close to the actual physical state. This method optimizes the model through multiple rounds of relaxation (energy optimization) and compression (adjusting the structural density) so that it is as close to the global energy minimum as possible at the end of the simulation, thereby improving the accuracy of the simulation. Through at least 30 rounds of annealing (i.e., multiple heating and cooling processes), it can be ensured that the final model structure is close to the global minimum in energy, making it more stable. This is because the annealing process can help the system escape the local minimum energy state and find a more stable global minimum energy structure.

[0124] Fig.10 is a flowchart of a random forest model training method according to an exemplary embodiment. Fig.10 As shown, the method includes:

[0125] S1001, setting a plurality of different groups of sample data, wherein the sample data includes the number of siloxane prepolymer structure models, the number of dimethylsiloxane monomer structure models, and at least part of the temperature and pressure parameters in the method described in steps S901 to S911.

[0126] S1002, respectively apply each set of sample data to the method described in step S901 to step S911 and execute the method, and after the method is executed, obtain multiple sets of lengths, widths and heights of the three-dimensional box.

[0127] S1003, calculating the degree of crosslinking based on the length, width and height using the following formula:

[0128]

[0129] Wherein, CKI represents the degree of cross-linking, X, Y and Z are the length, width and height of the three-dimensional box, respectively, and n is the number of siloxane prepolymer structure models.

[0130] S1004, the cross-linking degree and corresponding sample data are combined into a training data set, and a random forest model is trained based on the training data set to obtain a trained random forest model for predicting the cross-linking degree.

[0131] In step S1001, the number of siloxane prepolymer structure models, the number of dimethylsiloxane monomer structure models, and at least part of the temperature and pressure parameters in the method described in steps S901 to S911 can be composed into sample data, and the data value of at least one item can be adjusted to obtain multiple groups of sample data. For example, all temperature and pressure parameters used in the calculation process of steps S902 to S911, as well as the number of siloxane prepolymer structure models and the number of dimethylsiloxane monomer structure models can be composed into a group of sample data, and one or more data can be adjusted to obtain multiple groups of sample data. Of course, when determining the temperature and pressure parameters in steps S906 to S911, only the temperature and pressure parameters used in the calculation process of steps S902 to S905 can be used to compose the sample data.

[0132] After step S1001 is completed, step S1002 may be executed. In step S1002, each set of parameters in the sample data is used to set Fig. 9 The corresponding steps in the method are performed after the application. After the execution is completed, the size of the three-dimensional box changes, and the length, width and height of the three-dimensional box are obtained after each execution to obtain multiple sets of size data.

[0133] After step S1002 is completed, step S1003 may be executed. In step S1003, the crosslinking degree of the dimethylsiloxane dense membrane structure is calculated based on the length, width and height in step S1002. Specifically, the calculation may be performed using the following formula:

[0134]

[0135] Among them, CKI represents the degree of crosslinking, X, Y and Z are the length, width and height of the three-dimensional box respectively, and n is the number of siloxane prepolymer structure models. Through this formula, each new prepolymer is regarded as a crosslinking site, and the degree of crosslinking is defined as the number of crosslinking sites per cubic centimeter (mmol / cm³).

[0136] After step S1003 is completed, step S1004 can be executed. In step S1004, the crosslinking degree and the corresponding sample data are combined into a training data set. For example, after applying the first group of sample data to obtain the first crosslinking degree data, the sample data and the crosslinking degree data are combined into a group of training data, and then the crosslinking degrees of other sample data are obtained in sequence, and the sample data and the corresponding crosslinking degrees are combined into multiple groups of training data to obtain a training data set. Then, the training data set is used to train the random forest model. Since the data in the training data set is a continuous value, the output is also a continuous value.

[0137] Specifically, in step S1004, the training data set can be cleaned first, and a box plot or z-score method can be used to clean the training data set, identify outliers, remove invalid or abnormal data points, and improve data quality. In the case of a small number of training data sets, data enhancement can be performed by, for example, adversarial neural networks to expand the number of training data sets for training. Of the training data sets obtained by preprocessing, 80% of the training data is used as a training set for training a random forest model, and 20% is used as a test set for testing the training results. The training set can then be used to train a random forest model.

[0138] During training, you can first extract samples from the training data set multiple times, and each time use the bootstrap sampling method to extract training samples from the training data set to obtain multiple groups of training samples, each group of training samples is a subset of the training set. For this random forest model, the number of decision trees can be set to any value between 100-200, for example, the number of decision trees is selected to be 200, the maximum number of features of each decision tree is any value between 4-6, for example, it can be 6, and the maximum depth of the tree is 10. Generate a decision tree through each group of training samples to obtain multiple decision trees, and finally calculate the average value of the output of each decision tree to obtain the prediction result of the final random forest model.

[0139] After the random forest model is trained, it can be applied to the training set, algorithm regression can be performed, and then the prediction test can be performed on the test set. Specifically, the mean square error (MSE), root mean square error (RMSE), mean absolute error (MAE), and R 2 The smaller the first three items, the higher the R 2 The closer the coefficient of determination is to 1, the better the performance of the trained model.

[0140] Through the above technical solution, the relevant parameters for constructing a dimethylsiloxane dense film are used as sample data, and the length, width and height of the three-dimensional box obtained after the calculation are used to calculate the cross-linking degree. Based on the cross-linking degree and the corresponding sample data to form a training data set, a random forest model is trained to obtain a prediction model with good robustness and high accuracy, and there is no need to normalize the input data. The trained random forest model can predict the cross-linking degree in advance without building a model, reducing the resource consumption of molecular dynamics simulation calculations, saving time and computing power, and improving R&D efficiency.

[0141] Fig.11 is a flowchart of a method for predicting crosslinking degree based on artificial intelligence according to an exemplary embodiment, see Fig.11 , the method comprising:

[0142] S1101, obtaining temperature, pressure, number of siloxane prepolymer structure models, and number of dimethylsiloxane monomer structure models;

[0143] S1102, input the acquired temperature, pressure, number of siloxane prepolymer structure models and number of dimethylsiloxane monomer structure models into the random forest model trained in steps S1001 to S1004, and obtain the predicted output of the crosslinking degree of the dimethylsiloxane dense membrane structure.

[0144] In step S1101, the temperature and pressure parameters of the molecular dynamics simulation calculation of the dimethylsiloxane dense film structure to be predicted, as well as the number of siloxane prepolymer structure models and the number of dimethylsiloxane monomer structure models can be obtained. Then, step S1102 is entered to input these data into the random forest model trained in advance from step S1001 to step S1004 to obtain the predicted output cross-linking degree as the prediction result, which can be used as a reference for the corresponding experiment. If the predicted cross-linking degree does not meet expectations, the corresponding parameters can be changed and predicted again.

[0145] Through the above technical solution, the trained random forest model can quickly predict the cross-linking degree of dimethylsiloxane dense membrane structure calculated by molecular dynamics simulation under different conditions, avoiding the need to re-run time-consuming molecular dynamics simulation each time, reducing the dependence on molecular dynamics simulation calculations and its computing power and time resource consumption. This method is particularly suitable for scenarios where experimental conditions need to be adjusted multiple times.

[0146] Fig.12 FIG. 1 is a block diagram of an electronic device according to an exemplary embodiment. Fig.12As shown, the electronic device 700 may include: a processor 701 , a memory 702 . The electronic device 700 may also include one or more of a multimedia component 703 , an input / output (I / O) interface 704 , and a communication component 705 .

[0147] The processor 701 is used to control the overall operation of the electronic device 700 to complete all or part of the steps in the above-mentioned method for constructing a dense polydimethylsiloxane film structure, the random forest model training method, or the cross-linking degree prediction method based on artificial intelligence. The memory 702 is used to store various types of data to support the operation of the electronic device 700, which may include, for example, instructions for any application or method operating on the electronic device 700, and application-related data, such as contact data, sent and received messages, pictures, audio, video, etc. The memory 702 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, disk or optical disk. The multimedia component 703 may include a screen and an audio component. The screen may be, for example, a touch screen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signal may be further stored in the memory 702 or sent via the communication component 705. The audio component also includes at least one speaker for outputting audio signals. The I / O interface 704 provides an interface between the processor 701 and other interface modules, and the above-mentioned other interface modules can be a keyboard, a mouse, buttons, etc. These buttons can be virtual buttons or physical buttons. The communication component 705 is used for wired or wireless communication between the electronic device 700 and other devices. Wireless communication, such as Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, 4G, NB-IOT, eMTC, or other 5G, etc., or a combination of one or more of them, is not limited here. Therefore, the corresponding communication component 705 may include: Wi-Fi module, Bluetooth module, NFC module, etc.

[0148] In an exemplary embodiment, the electronic device 700 can be implemented by one or more application specific integrated circuits (ASIC), digital signal processors (DSP), digital signal processing devices (DSPD), programmable logic devices (PLD), field programmable gate arrays (FPGA), controllers, microcontrollers, microprocessors or other electronic components to execute the above-mentioned method for constructing a dense polydimethylsiloxane film structure, the random forest model training method or the cross-linking degree prediction method based on artificial intelligence.

[0149] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided, and when the program instructions are executed by a processor, the steps of the method for constructing a dense polydimethylsiloxane film structure, the random forest model training method, or the cross-linking degree prediction method based on artificial intelligence are implemented. For example, the computer-readable storage medium can be the memory 702 including the program instructions, and the program instructions can be executed by the processor 701 of the electronic device 700 to complete the method for constructing a dense polydimethylsiloxane film structure, the random forest model training method, or the cross-linking degree prediction method based on artificial intelligence.

[0150] The preferred embodiments of the present disclosure are described in detail above in conjunction with the accompanying drawings; however, the present disclosure is not limited to the specific details in the above embodiments. Within the technical concept of the present disclosure, a variety of simple modifications can be made to the technical solution of the present disclosure, and these simple modifications all fall within the protection scope of the present disclosure.

[0151] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. In order to avoid unnecessary repetition, the present disclosure will not further describe various possible combinations.

[0152] In addition, various embodiments of the present disclosure may be arbitrarily combined, and as long as they do not violate the concept of the present disclosure, they should also be regarded as the contents disclosed by the present disclosure.

Claims

1. A method for constructing a dense polydimethylsiloxane film structure, characterized in that: The method comprises: Constructing the structural model of siloxane prepolymer and dimethylsiloxane monomer; Performing geometric optimization on the siloxane prepolymer structure model and the dimethylsiloxane monomer structure model; A first number of geometrically optimized siloxane prepolymer structure models and a second number of geometrically optimized dimethylsiloxane monomer structure models are randomly placed in a three-dimensional box of a preset size for random mixing to obtain an amorphous mixed system; Using a conjugate gradient algorithm to adjust the atomic coordinates of the amorphous mixed system, and find a stable structure corresponding to the minimum static energy of the amorphous mixed system; For the stable structure, under a preset ReaxFF force field, based on a preset Verlet-Velocity algorithm for integrating the equation of motion and a preset Nose-Hoover algorithm for adjusting the temperature, a temperature equilibrium simulation is performed at any temperature between 500K and 800K, and then a temperature equilibrium simulation is performed at any temperature between 298K and 330K, so that the siloxane prepolymer structure model in the stable structure is connected to the silicon atoms and oxygen atoms of the dimethylsiloxane monomer structure model through its own oxygen active sites and silicon active sites, respectively, to form a dimethylsiloxane dense film structure.

2. The method according to claim 1, characterized in that: The method of performing a temperature balance simulation at any temperature between 500K and 800K based on a preset Verlet-Velocity algorithm for integrating the motion equation and a preset Nose-Hoover algorithm for adjusting the temperature, and then performing a temperature balance simulation at any temperature between 298K and 330K, includes: Based on the Verlet-Velocity algorithm with a preset time step of 0.1 fs for integration of the equations of motion and the Nose-Hoover algorithm with a preset standard relaxation time of 10 fs for adjusting the temperature, a temperature equilibrium simulation was performed at 500 K for 50 ps and then at 300 K for 100 ps.

3. The method according to claim 1, characterized in that The construction of the siloxane prepolymer structure model and the dimethylsiloxane monomer structure model comprises: Constructing a dimethylsiloxane monomer structure model and a tetraethyl silicate structure model, and connecting the oxygen atom of one dimethylsiloxane monomer structure model with the silicon atom of another dimethylsiloxane monomer structure model to obtain a dimethylsiloxane dimer model; Two ethoxy groups of the four ethoxy groups of the tetraethyl silicate structural model are removed, and the ethyl groups of the other two ethoxy groups are removed to obtain an intermediate molecular model, and the silicon atom of the intermediate molecular model is connected to the oxygen atoms containing unsaturated bonds of two dimethylsiloxane dimer models, and the two oxygen atoms of the intermediate molecular model are respectively connected to the silicon atom containing unsaturated bonds of a dimethylsiloxane dimer model to obtain a siloxane prepolymer structural model.

4. The method according to claim 1, characterized in that: The siloxane prepolymer structure model and the dimethylsiloxane monomer structure model are geometrically optimized, including: In Materials Studio software, the Geometry optimization task in the Forcite module was used, the Smart algorithm was adopted, the Energy parameter was set to 0.001 kcal / mol, the Force parameter was set to 0.5 kcal / mol / Å, the Displacement parameter was set to 0.015 Å, the maximum number of iterations was set to 500 steps, the force field used was COMPASSIII force field, the charge parameter was set to Forcefield assigned, the charge group was set to automatic force field assignment, the electrostatic term was set to Ewald summation method, the accuracy parameter was set to 0.001 kcal / mol, the van der Waals term was set to Atom-based summation method, the cutoff method used was Cubic spline, the cutoff radius was set to 12.5 Å, the bond width was set to 1Å, and Long range correction was used for geometry optimization during the calculation process.

5. The method according to claim 1, characterized in that The second number is any multiple between 8 and 40 times of the first number.

6. The method according to claim 5, characterized in that The first number is 30, and the second number is 274.

7. The method according to claim 1, characterized in that The method further comprises: Select the NPT ensemble, set the pressure to 1000 bar, the temperature to 300 K, and the simulation time to 300 ps, ​​and perform a volume compression simulation calculation based on molecular dynamics simulation on the dimethylsiloxane dense membrane structure; Select the NVT ensemble, set the temperature to 800K, and the simulation time to 100ps, and perform a heating relaxation simulation calculation based on molecular dynamics simulation on the dimethylsiloxane dense membrane structure that has undergone volume compression; Select the NVT ensemble, set the temperature to 300K, and the simulation time to 100ps, and perform a cooling simulation calculation based on molecular dynamics simulation on the dimethylsiloxane dense membrane structure after heating and relaxation; Select the NPT ensemble, set the pressure to 100 bar, the temperature to 300 K, and the simulation time to 300 ps, ​​and perform a recompression simulation calculation based on the molecular dynamics simulation on the dimethylsiloxane dense membrane structure that has undergone the cooling simulation calculation; When the number of iterative calculations is less than the threshold and the density difference of the dimethylsiloxane dense film obtained by two adjacent recompression simulations is greater than 0.001 g / cm 3 , returning to the step of selecting the NPT ensemble, setting the pressure to 1000 bar, the temperature to 300 K, the simulation time to 300 ps, ​​and performing a volume compression simulation calculation based on molecular dynamics simulation on the dimethylsiloxane dense membrane structure; When the number of iterative calculations is greater than or equal to the threshold or the density difference of the dimethylsiloxane dense film obtained by two adjacent recompression simulations is less than or equal to 0.001 g / cm 3 When the NPT ensemble is selected, the pressure is set to 1 bar, the temperature is set to 300 K, and the simulation time is set to 10 ns. The molecular dynamics simulation calculation is performed on the dimethylsiloxane dense membrane structure obtained by the final compression simulation calculation to obtain a stable dimethylsiloxane dense membrane structure.

8. A random forest model training method, characterized in that: The method comprises: Setting a plurality of different sets of sample data, wherein the sample data includes the number of siloxane prepolymer structure models, the number of dimethylsiloxane monomer structure models, and at least part of the temperature and pressure parameters in the method according to claim 7; Applying each set of sample data to the method according to claim 7 and executing the method, and obtaining multiple sets of lengths, widths and heights of the three-dimensional box after the method is executed; The degree of crosslinking is calculated based on the length, width and height using the following formula: ； Wherein, CKI represents the degree of cross-linking, X, Y and Z are the length, width and height of the three-dimensional box, respectively, and n is the number of siloxane prepolymer structure models; The cross-linking degree and corresponding sample data form a training data set, and a random forest model is trained based on the training data set to obtain a trained random forest model for predicting the cross-linking degree.

9. A method for predicting crosslinking degree based on artificial intelligence, characterized in that: The method comprises: Obtaining temperature, pressure, number of siloxane prepolymer structure models, and number of dimethylsiloxane monomer structure models; The obtained temperature, pressure, number of siloxane prepolymer structure models and number of dimethylsiloxane monomer structure models are input into the random forest model trained by the method according to claim 8 to obtain the predicted output of the crosslinking degree of the dimethylsiloxane dense film structure.

10. An electronic device, characterized in that: include: a memory having a computer program stored thereon; A processor, configured to execute the computer program in the memory to implement the steps of the method according to any one of claims 1 to 9.

Images