Search

CN-122024927-A - Polymer compound conductive network structure screening method, system and device

CN122024927ACN 122024927 ACN122024927 ACN 122024927ACN-122024927-A

Abstract

The invention discloses a method, a system and a device for screening a high polymer composite conductive network structure, and relates to the technical field of conductive high polymer nano composite materials, wherein the method comprises the steps of obtaining molecular dynamics simulation track data of high polymer composite systems of carbon nano tubes with different concentrations, sequentially encoding particle nodes, performing independent thermal encoding according to particle types and belonged molecular IDs, generating a node attribute matrix, and generating an adjacent matrix according to a bonding relation among particles to construct a corresponding graph structure data set; inputting the graph structure data set into a prediction model based on a graph attention network, and training by adopting an incremental learning method; according to the trained prediction model, the optimal high polymer compound conductive network structure is screened out, and the method can capture complex multi-stage structure information and better predict the property of the polymer affected by the complex multi-stage structure information.

Inventors

  • MAO JIASHUN

Assignees

  • 西南医科大学

Dates

Publication Date
20260512
Application Date
20260128

Claims (8)

  1. 1. The screening method of the high polymer compound conductive network structure is characterized by comprising the following steps of: Acquiring molecular dynamics simulation track data of a polymer compound system of carbon nanotubes with different concentrations, and carrying out normalization processing on particle three-dimensional coordinates of each frame in the molecular dynamics simulation track data; Sequentially encoding the particle nodes based on the normalized particle three-dimensional coordinates, and performing single-heat encoding according to the particle types and the belonged molecular IDs to generate a node attribute matrix, and generating an adjacent matrix according to the bonding relation among the particles to construct a corresponding graph structure data set; taking a node attribute matrix and an adjacent matrix as input and taking the overall conductivity corresponding to each frame of graph structure as output to construct a prediction model based on a graph attention network; inputting the graph structure data set into a prediction model for training according to the sequence from low concentration to high concentration by adopting an incremental learning strategy, wherein the model parameters obtained by previous concentration training are used as initial parameters of next higher concentration data training, and iteration is sequentially carried out until the training of all concentration data is completed; the method comprises the steps of conducting conductivity prediction on a high polymer compound conducting network structure to be screened by using a trained prediction model, extracting an attention score matrix of the network structure, quantitatively analyzing connectivity of the high polymer compound conducting network structure to be screened based on the attention score matrix, and screening out an optimal high polymer compound conducting network structure according to the predicted conductivity and connectivity analysis result.
  2. 2. The method for screening a conductive network structure of a polymer composite according to claim 1, wherein the overall conductivity corresponding to each frame structure is calculated by a macroscopic resistance method.
  3. 3. The method for screening an optimal polymer composite conductive network structure according to claim 1, wherein the calculation process of the CNT concentration of the carbon nanotubes is as follows: CNT concentration = 。
  4. 4. The method according to claim 1, wherein the prediction model based on graph attention network comprises a residual connection module, a multi-head attention mechanism and a global pooling strategy integrating differential pooling.
  5. 5. The method for screening the conductive network structure of the polymer composite according to claim 1, wherein the method for quantitatively analyzing connectivity of the conductive network structure of the polymer composite to be screened based on the attention score matrix is characterized by specifically comprising the steps of reconstructing a network by taking an N side before ranking of attention scores as a connection threshold, calculating a topology index of the reconstructed network, and analyzing the connectivity according to the topology index, wherein the topology index comprises average degree, connected components, global efficiency, clustering coefficient, network density and shortest path length.
  6. 6. A polymer composite conductive network structure screening system, comprising: The acquisition module is used for acquiring molecular dynamics simulation track data of the polymer compound systems of the carbon nanotubes with different concentrations, and carrying out normalization processing on the particle three-dimensional coordinates of each frame in the molecular dynamics simulation track data; The data set construction module is used for sequentially encoding the particle nodes based on the normalized particle three-dimensional coordinates, performing independent thermal encoding according to the particle types and the belonged molecular IDs, generating a node attribute matrix, and generating an adjacent matrix according to the bonding relation among the particles so as to construct a corresponding graph structure data set; The model training module is used for inputting the graph structure data set into the prediction model from low concentration to high concentration by adopting an incremental learning strategy, and specifically comprises the steps of sequentially iterating by taking model parameters obtained by previous concentration training as initial parameters of next higher concentration data training until the training of all concentration data is completed; The screening module is used for conducting conductivity prediction on the polymer compound conductive network structure to be screened by utilizing the trained prediction model, extracting an attention score matrix of the network structure, quantitatively analyzing connectivity of the polymer compound conductive network structure to be screened based on the attention score matrix, and screening out the optimal polymer compound conductive network structure according to the predicted conductivity and the connectivity analysis result.
  7. 7. The screening computer device for the conductive network structure of the polymer compound is characterized by comprising a memory, a processor and a computer program stored in the memory, wherein the processor realizes the steps of the screening method for the conductive network structure of the polymer compound according to any one of claims 1-5 when executing the computer program.
  8. 8. A readable storage medium, wherein the readable storage medium stores a computer program comprising program instructions for performing the steps of the polymer composite conductive network structure screening method of any one of claims 1-5 when executed by a processor.

Description

Polymer compound conductive network structure screening method, system and device Technical Field The invention relates to the technical field of conductive polymer nano composite materials, in particular to a method, a system and a device for screening a conductive network structure of a polymer composite. Background With the continuous progress of technology, research and development of new materials have become key forces for promoting the development of modern technology. In particular, in the fields of electronics, energy, medical treatment, etc., there is an increasing demand for high-performance materials. Conductive polymer nanocomposites (Conductive Polymer Nanocomposites, CPNs) have received great attention due to their unique physical, chemical, and electrical properties. In recent years, scientists have made many studies on self-assembly, phase structure and electrical properties of CPNs composite systems using experimental methods. But most are based on experimental methods. The electrical properties of CPNs are controlled by one trial and error. However, CPNs has a huge parameter space of the composite architecture, so that huge economic and time costs are required for determining the optimized synthesis conditions, and the method is limited by the technical means of a detection instrument, so that the experimental method is difficult to observe the local conformation of a high molecular chain and the dynamic evolution process of the local conformation on the molecular scale in real time. At present, a feature extraction method based on SMILES character strings and monomer proportions is proposed, wherein monomers are formed into homopolymers through a weighted chain, and the Mooney fingerprint and the Molde are predicted by gathering information from adjacent nodes and edges in a polymer chain through a GCN method, so that the glass transition temperature (Tg) and the thermal decomposition temperature (Td) are further predicted. However, due to the influence of complex polydisperse, multiscale and multilevel complex structures of high polymers, the method cannot capture complex multilevel structure information, and the property of the polymer influenced by the complex multilevel structure cannot be predicted easily. Disclosure of Invention Aiming at the defect that the prior art can not capture complex multilevel structure information and is difficult to predict the property of the polymer affected by the complex multilevel structure information, the invention provides a method, a system and a device for screening a high polymer compound conductive network structure, which directly learn physical rules from three-dimensional structure data by adopting an incremental graph neural network model, the mechanism of influence of conductive nanoparticle networks with different concentrations on the electrical performance of CPNs under the same polymer matrix (homopolymer system) is researched, so that the structure-activity relationship between the microcomponents of CPNs and the electrical performance is further clarified, and the problems in the prior art are solved. A screening method of a high molecular compound conductive network structure comprises the following steps: Acquiring molecular dynamics simulation track data of a polymer compound system of carbon nanotubes with different concentrations, and carrying out normalization processing on particle three-dimensional coordinates of each frame in the molecular dynamics simulation track data; Sequentially encoding the particle nodes based on the normalized particle three-dimensional coordinates, and performing single-heat encoding according to the particle types and the belonged molecular IDs to generate a node attribute matrix, and generating an adjacent matrix according to the bonding relation among the particles to construct a corresponding graph structure data set; taking a node attribute matrix and an adjacent matrix as input and taking the overall conductivity corresponding to each frame of graph structure as output to construct a prediction model based on a graph attention network; inputting the graph structure data set into a prediction model for training according to the sequence from low concentration to high concentration by adopting an incremental learning strategy, wherein the model parameters obtained by previous concentration training are used as initial parameters of next higher concentration data training, and iteration is sequentially carried out until the training of all concentration data is completed; the method comprises the steps of conducting conductivity prediction on a high polymer compound conducting network structure to be screened by using a trained prediction model, extracting an attention score matrix of the network structure, quantitatively analyzing connectivity of the high polymer compound conducting network structure to be screened based on the attention score matrix, and screening out an optimal high polymer