Method and System for Predicting Glass Transition Temperature of Polymer

Abstract

The present invention provides a computer-implemented method for predicting the glass transition temperature of a polymer from the chemical structure of the polymer. The present invention provides a method for graphically representing the chemical structure of a polymer, and machine-learns the chemical structure of the polymer based on information defining the connection relationship between each atom constituting the repeating unit of the polymer and the connection nodes to which the repeated repeating units are connected, and calculates the glass transition temperature of the polymer from the graph information of the polymer. In addition to the chemical structure of the polymer, the present invention provides a method and system for calculating the glass transition temperature of the polymer while also reflecting the molecular weight information.

Description

Technical Field

[0001] The present invention relates to a method and a system for predicting the glass transition temperature of a polymer using an artificial intelligence neural network.

Background Art

[0002] In order to input the structure of a compound into a computer device and perform predetermined processing, various methods have been proposed to represent the compound structure in a form recognizable by the computer device. Typically, there are SMILES that represent the structure of a compound as a character string, or a method of representing a character string converted to SMILES by descriptors.

[0003] However, for polymers that are not single molecules, it is difficult to represent them using the SMILES method, so BigSMILES and hierarchical descriptors have been proposed.

[0004] However, such conventional technologies have the problems that as the molecule becomes larger, the representation method becomes increasingly complex and only represents the polymer fragmentarily, there is no technology capable of processing molecules based on the representation, and it is impossible to accurately represent the characteristics of a polymer in which the repeating unit is infinitely repeated.

[0005] On the other hand, when predicting a predetermined property of a polymer, conventionally, as disclosed in the following non-patent documents, the predetermined property of the polymer has been predicted by reflecting the structural property of the polymer.

[0006] However, among the properties of polymers, the glass transition temperature (Tg) is affected not only by the structure of the polymer but also by the molecular weight of the polymerized polymer. Therefore, there is a disadvantage that the accurate glass transition temperature cannot be predicted by the conventional technology.

Prior Art Documents

Patent Documents

[0007] [Patent Document 1] Republic of Korea Published Patent Gazette No. 2021-0042777 [Patent Document 2] Republic of Korea Published Patent Gazette No. 2021-0110539 [Non-Patent Document]

[0008] [Non-Patent Document 1] Machine-learning predictions of polymer properties with Polymer Genome, Journal of Applied Physics 128, 171104, 2020 [Non-Patent Document 2] Machine learning discovery of high-temperature polymers, Matter, Volume 4, Issue 5, 5 May 2021, pages 1454-1456 [Summary of the Invention] [Problems to be Solved by the Invention]

[0009] Therefore, in order to solve the above-mentioned problems, the present invention provides a data structure for representing a polymer substance by graphic information, and further predicts the glass transition temperature of a polymer using the data structure representing the graphic information of the polymer, and its purpose is to provide a method and a system for predicting while reflecting the molecular weight element of the polymer. [Means for Solving the Problems]

[0010] To solve the above problems, the present invention provides a computer-implemented method for graphically describing the chemical structure of a selected polymer and calculating a predicted glass transition temperature of the polymer, the method comprising: a step of obtaining a polymer glass transition temperature characteristic information prediction artificial neural network that is machine-learned to predict the glass transition temperature characteristic information of the polymer based at least in part on chemical structure data related to the selected polymer; a step of obtaining polymer graph information that graphically describes the chemical structure of the selected polymer as graphic data; a step of inputting the obtained polymer graph information into the graph neural network; a step of obtaining the polymer glass transition temperature characteristic information of the selected polymer from the output of the graph neural network; and a step of calculating the glass transition temperature of the selected polymer by substituting the molecular weight information of the selected polymer into the polymer glass transition temperature characteristic information.

[0011] At this time, the polymer graph information that graphically describes the chemical structure of the selected polymer includes at least one of: polymer graph node connection information, which represents the connection relationship between nodes corresponding to each atom constituting the repeating unit forming the polymer and connection nodes connected at connection points where the repeating unit is repeatedly connected; polymer node attribute information, which represents the attribute information of the nodes corresponding to each atom constituting the repeating unit forming the polymer and the attributes of the connection nodes connected at connection points where the repeating unit is repeatedly connected; and polymer edge attribute information, which represents the attributes of edges connecting the nodes corresponding to each atom constituting the repeating unit forming the polymer to each other and connection edges connecting at least one of the nodes to the connection nodes.

[0012] The present invention also provides a polymer glass transition temperature calculation system, which includes a data input unit that receives input of polymer graph information describing the chemical structure of a selected polymer in graphical data and molecular weight data of the polymer, an artificial neural network model unit that predicts and calculates glass transition temperature characteristic information of the polymer from the polymer graph information, and a glass transition temperature calculation unit that calculates the glass transition temperature of the polymer from the glass transition temperature characteristic information of the polymer and the molecular weight data of the polymer. At this time, the glass transition temperature characteristic information of the polymer is the T value and K in the following formula (1), and the molecular weight data of the polymer is M in the following formula. The glass transition temperature calculation unit calculates the glass transition temperature T from the following formula. g,inf value and K, and the molecular weight data of the polymer is M in the following formula. n The glass transition temperature calculation unit calculates the glass transition temperature T from the following formula. g

[0013]

Equation

[0014] (T g is the glass transition temperature of a polymer having a molecular weight of M n , T g,inf is the glass transition temperature when the molecular weight is infinite, K is a polymer free volume related variable, and M n is the number average molecular weight of the polymer.)

Advantages of the Invention

[0015] ​According to the present invention, by providing a data structure capable of representing a polymer by graphic information that can be used by a computer, it is possible to realize a method and a system for predicting the characteristic information of a polymer substance by machine learning the relationship between the molecular structure and the characteristic information of the polymer substance using a graph neural network (GNN) and a message passing neural network (MPNN). Furthermore, when predicting the glass transition temperature of a polymer, by reflecting the molecular weight factor, it has become possible to provide a method and a system for predicting the glass transition temperature of polymer characteristic information with improved accuracy compared to the conventional technology.

[0016] The drawings attached to this specification illustrate preferred embodiments of the present invention and are for the purpose of further understanding the technical idea of the present invention together with the content of the invention. Therefore, the present invention is not to be construed as being limited only to the matters described in the drawings.

Brief Description of the Drawings

[0017]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Embodiments for Carrying Out the Invention

[0018] 1. Method for graphically representing polymers according to the present invention

[0019] First, a method for graphically representing a polymer according to the present invention will be described based on FIGS. 1 and 2.

[0020] (a) and (b) of FIG. 1 are diagrams showing different polymers and the structures of the monomers forming each polymer. However, when attempting to process molecular structure data using an artificial neural network, particularly when trying to represent the molecular structure as a graph for application to, for example, a GNN, there has been no method for effectively representing the chemical structure of a polymer as a graph.

[0021] Although the monomers forming the polymers shown in (a) and (b) of FIG. 1 are different from each other, they are the same except for a part of the ends of the monomers. Therefore, their monomer graphs are represented as being very similar. Thus, it was inaccurate to represent the polymer graph simply as the monomer graph and represent the polymer graph as a repetition of the monomer graph.

[0022] Therefore, in the present invention, a novel method for graphically representing a polymer as shown in FIG. 2 is proposed. Hereinafter, based on FIG. 2, the method for graphically representing polymer information according to the present invention will be described.

[0023] First, in the graph representation method of the polymer of the present invention, the monomers constituting the polymer and the repeating unit monomers that are repeated to constitute the polymer are derived. FIG. 1 shows an example of the representation of a conventional polymer and the display of the monomers constituting the polymer, and FIG. 2 is a diagram showing the repeating unit monomers constituting the polymer of FIG. 1. In the repeating unit monomers of FIG. 2, for the polymer representation method of the present invention, each atom is represented by a node 10, and the bond between the atoms is represented by an edge 20. Next, when the repeating unit monomers repeatedly form a polymer, which is one of the features of the present invention, the location where the repeating unit monomers are repeatedly connected is represented by an attachment point 30, and the node to which the attachment point is connected is represented as the "connection node". Further, the attachment point 30 virtually represents the connection nodes of adjacent repeating unit monomers that are repeatedly connected, and the connection connecting the attachment point and the connection node is represented by an attachment edge 40.

[0024] According to such a graph representation method of the polymer of the present invention, when learning an artificial neural network model based on the chemical structure of the polymer, particularly when applying an artificial neural network model such as an MPNN (message passing neural network) or a GNN (graph neural network) based on message passing to proceed with learning, in addition to the monomer structure information, learning can be carried out such that the information of the nodes on the opposite side where the monomers are connected is passed on with only the addition of minimal information, and learning that reflects the unit repeating characteristics of the polymer becomes possible.

[0025] 2. Data conversion of the graph representation of the polymer according to the present invention so that it can be used by a computer

[0026] The present invention converts the molecular structure or chemical structure of the above-described polymer into a data structure in a form that can be used for computer arithmetic processing, that is, "graph data", by means of a graph representation method for graphically representing the chemical structure of the polymer. In the present invention, this is referred to as "polymer graph data".

[0027] The polymer graph data according to the present invention, which describes the chemical structure of a predetermined or selected polymer by means of graphic information, includes at least one of node 10 representing two or more nodes representing each atom constituting the repeating unit of the monomer forming the polymer, node information which is data related to node 10, edge information (edge data) which is data related to one or more edges 20 representing the bonds between the respective nodes, attaching node information (attaching node data) which is data related to at least one attaching node which is a location where the repeating unit forming the polymer is repeatedly connected, and attaching edge information (attach edge data) which is data related to attach edge 40 which is a connection connecting the attachment point 30 and the attaching node.

[0028] The difference between the monomer graph data and the polymer graph data of the present invention is that in the expression of the graph information of the repeating unit including the monomer graph data, the polymer graph data of the present invention represents the node 10 connected to the attachment point 30 and its edge information by the attaching node and the attaching edge.

[0029] 2.1. Adjacency Matrix and Feature Matrix

[0030] When digitizing the graph structure of a monomer so that it can be used by a computer, there can be various data representation methods. Usually, the edge information and node information can be represented by an adjacency matrix and a characteristic matrix, respectively. The edge information is information related to the bonds between atoms, and the node information is information related to each atom.

[0031] According to the polymer representation method of the present invention, in addition to the conventional monomer graph representation method, it is possible to further add information on the nodes and connection edges 40 connected to the connection point 30 and convert the polymer graph into data recognizable by a computer.

[0032] (1) Representation of the adjacency matrix of the monomer graph

[0033] For the sake of simplicity of explanation, taking a simple structure as shown in FIG. 3 as an example, the adjacency matrix A of the monomer repeating unit composed of nodes 1, 2, 3, and 4 can be represented as follows.

[0034]

Equation

[0035] The adjacency matrix A has rows and columns corresponding to the number of nodes, and each component A- i,j represents information on whether the i-th (i = 1,..., 4) node and the j-th (j = 1,..., 4) node are connected. The value of each matrix component indicates the connection information with other nodes. For example, since node 1 is only connected to node 2 and not connected to nodes 1, 3, and 4, A of the adjacency matrix 1,j (j = 1,..., 4) = [0 1 0 0].

[0036] (2) Representation of the adjacency matrix of the polymer graph according to the present invention

[0037] According to the polymer graph information representation method of the present invention, the polymer adjacency matrix PA (polymer Adjacence matrix), which is the representation of the adjacency matrix of the polymer graph associated with the example in FIG. 3, can be represented as follows.

[0038]

Number

[0039] As shown in FIG. 3, for the polymer adjacency matrix PA, two connection points 5 and 6 are added, which respectively represent the nodes 4 and 1 of the adjacent monomers. Therefore, the nodes 1 and 4 are represented as being connected to the nodes 4 and 1 respectively. That is, in the monomer, since there is no node to which the nodes 1 and 4 are connected, the matrix value indicating the connection relationship between the node 1 and the node 4 is A 1,4 = 0, A 4,1 = 0. However, in the polymer representation method of the present invention, since the nodes 1 and 4 are respectively connected to the nodes 4 and 1 of the adjacent monomers, PA of the polymer adjacency matrix PA 1,4 = 1, PA 4,1 = 1, which indicates that it includes the repetitive connection information of the polymer. That is, in the monomer, the nodes 1 and 4 that were not connected are connected to each other in the polymer. Therefore, in the polymer adjacency matrix that displays the connection edge information, they are represented as being connected to each other.

[0040] However, in this case, since the nodes 1 and 4 are not connected to each other within the monomer, in order to represent that it is not the connection within the monomer but the connection between the nodes of the adjacent repeating units, the node connection relationship can be represented by a negative number to represent the polymer adjacency matrix as follows, or the "connected node" characteristic connected to the connection point can be given to the characteristic of the node in the node characteristic matrix described later, or the adjacent repeating unit connection attribute can be given to the edge characteristic matrix to distinguish them.

[0041]

Number

[0042] To explain again, the graph information that graphs the chemical structure of the polymer according to the present invention and describes it with graphic data includes, in addition to the graph information of the monomer repeating unit, connection node information representing a connection node that is a node to which monomers are repeatedly connected, and information on a connection edge 40 connected to the connection node.

[0043] According to the representation method of the polymer adjacency matrix PA, it means that the monomer repeating unit composed of nodes 1, 2, 3, and 4 is repeatedly connected to the nodes 4 and 1 of the adjacent repeating units at nodes 1 and 4, respectively. However, at this time, nodes 1 and 4 in FIG. 3 are not connected to each other within the monomer, but are connected to the nodes of adjacent repeating units. Therefore, these nodes are designated as "connection nodes", and the connection node attributes can be given to the node characteristic matrix, or the adjacent repeating unit connection attributes can be given to the edge characteristic matrix for distinction.

[0044] This can be achieved by using a polymer edge variable setting step of setting the interconnection relationship between nodes representing each atom constituting the monomer repeating unit of the selected polymer and connection nodes that are the locations where the monomers are repeatedly connected as a polymer connection edge variable, and a polymer graph edge information assignment step of assigning an attribute value of the interconnection relationship between the nodes and the connection nodes to the set polymer edge variable. As is clear from the above polymer adjacency matrix PA, the attribute value of the interconnection relationship between the nodes and the connection nodes can be determined as "1" and "0" indicating connection / non-connection, and this is represented by a matrix, which is the polymer adjacency matrix. The polymer adjacency matrix of the present invention has its component values, similar to a normal adjacency matrix, but is different in that, as described above, connection nodes are added to normal single-molecule graph nodes.

[0045] (3) Characteristic Matrix of Monomer Graph

[0046] In addition to the adjacency matrix indicating the connection relationship between nodes, the graph representation of a compound can have, as graph information, a characteristic matrix representing predetermined attribute information of each node and predetermined attribute information of each edge. In the exemplary monomer graph of FIG. 3, the monomer node characteristic matrix having attribute value information (exemplarily, three pieces of attribute information) of each atom can be represented as follows.

[0047]

Number

[0048] The node characteristic matrix NF can have rows corresponding to the number of each node and columns corresponding to the number of types of characteristic values to be represented for each node, and can represent a predetermined attribute to be represented by each atom. Each component NF i,j has the j-th characteristic value of the i-th node. The monomer characteristic matrix NF is described by taking, as an example, the case where each node has three arbitrary attribute values for the sake of explanation.

[0049] The information of the monomer graph can be represented by an edge characteristic matrix EF. Each row can represent each edge of the monomer, and each column data of each row can represent a predetermined attribute value of each edge. Examples of edge attribute values can include the type of chemical bond (single bond, double bond, etc.), the presence or absence of a ring, the presence or absence of conjugation, etc. For example, in the case of the monomer of FIG. 3, since there are three connecting edges, the edge characteristic matrix has three rows, and the edge attribute values to accommodate information are assigned to each column. An example of an edge characteristic matrix having three arbitrary edge attribute values is as follows.

[0050]

Number

[0051] (4) Characteristic Matrix of Polymer Graph According to the Present Invention

[0052] The polymer graph information according to the present invention can have a separate attribute value added to the attribute value of the monomer node characteristic matrix described above. For example, the polymer node characteristic matrix PF (Polymer Feature matrix), which is an exemplary polymer graph characteristic matrix in FIG. 3, can also have additional attribute information in the third column, as shown in the following example, and this can be represented by an attribute value indicating whether each node is a connected node. For example, the following polymer node characteristic matrix PNF can have connection node attribute information in the fourth column in addition to the monomer node characteristic matrix NF that accommodates three types of attribute information of the previous four nodes, as follows. In the example of FIG. 3, since nodes 1 and 4 are connected nodes each connected to an adjacent repeating unit, they have a value of "1", and the other nodes are represented as having a value of "0" in the fourth column.

[0053]

Number

[0054] This is achieved by using a polymer node variable setting step of setting, as polymer node variables, each atom of the monomer forming the selected polymer and the connection nodes that are the locations where the monomers are repeatedly connected, and a polymer graph node information assignment step of assigning connection nodes connected to the set polymer node variables and assigning node attribute values to each node. The node attribute values assigned to the nodes can include atomic number, hybridization (such as SP3, SP2), number of hydrogens, number of electrons, presence or absence of a ring, etc. In particular, as an embodiment of the present invention, it can have a connection node attribute value, which is an attribute value indicating the presence or absence of a node connected to an adjacent repeating unit value.

[0055] As another embodiment of the present invention, the connection node information can be represented by an edge property matrix. For example, like the previous polymer node property matrix, instead of or in addition to representing whether a node is a connection node by a node attribute value, an attribute value representing the presence or absence of a connection edge is represented in the edge property matrix. In the following exemplification representing (b) of FIG. 3, in order to accommodate the edges connecting connection nodes 1 and 4, that is, the information of edges 5_1 and 4_6 in (b) of FIG. 3, data in the 4th and 5th rows are added, and it can be seen that an attribute information column representing the presence or absence of a connection edge is added to the 4th column.

[0056]

Number

[0057] That is, according to the present invention, the method of representing the structure of a polymer by a graph and converting it into data that can be used by a computer can be selected from any one or more of the following methods.

[0058] 1) In an adjacency matrix representing the connection relationship of each node of a repeating unit composed of at least one monomer, assign the component of the "connection node" connected to the adjacent repeating unit as the "connection" attribute.

[0059] 2) Give a "connection presence / absence" attribute field to the node attribute matrix representing the attributes of each node of the repeating unit, and assign "connection" to the "connection presence / absence" attribute field of the "connection node".

[0060] 3) Add the edge connected to the connection node to the edge attribute matrix of the repeating unit as a connection edge, give a "connection presence / absence" attribute field to the edge attribute, and assign "connection" to the corresponding field of the connection edge.

[0061] 2.2. Method for converting the chemical structure of a polymer into data that can be processed by a computer

[0062] In the present invention, when learning an artificial neural network model described later and predicting the properties of a polymer using the learned artificial neural network, the chemical structure of the polymer must be digitized so that it can be processed by a computer. However, digitization is performed according to the method of the present invention as described below.

[0063] (1) Repeating unit information acquisition step

[0064] This is a step of acquiring repeating unit information, which is a unit in which monomers are repeated to form a polymer. In order to digitize connection node information in the repeating unit, which is one of the features of the present invention, repeating unit information forming a polymer is acquired. The repeating unit information is a step of confirming whether the repeating unit that is repeated to form a polymer is repeated as a single monomer or as a combination of single monomers, and describing the repeating unit in a graphical representation. The repeating unit can be a single monomer as shown in Fig. 2(a) or a case where monomers are connected as shown in Fig. 2(b).

[0065] (2) Polymer graph node connection information assignment step

[0066] In the present invention, the method of processing polymer graph information has a process of setting, as variables, the mutual connection relationships between nodes representing each atom constituting the repeating unit of the polymer selected as the analysis target and connection nodes connected to connection points, which are the locations where the repeating units are repeatedly connected.

[0067] The attribute values of the mutual connection relationships between the nodes and the connection nodes are assigned to the polymer node connection variables thus set. Assigning the attribute values of each connection relationship to the polymer node connection variables can utilize the polymer adjacency matrix cited as an example above.

[0068] (3) Polymer graph node attribute information assignment step

[0069] In addition, the polymer graph information processing method of the present invention can have a polymer node attribute variable setting step of setting, as polymer node attribute variables, the attribute information of the nodes corresponding to each atom constituting the repeating unit that forms the selected polymer and the attribute information of the connection nodes connected to the connection points, which are the locations where the repeating units are repeatedly connected, among the nodes.

[0070] Furthermore, it has a polymer graph node attribute information assignment step of assigning the attribute values of the nodes and the connection nodes to the polymer node attribute variables set in this way. The node attribute information has an attribute value indicating whether the node is a connection node or not. As the node attribute information of the connection node, unlike the nodes that are not connection nodes, a connection node attribute value indicating that it is connected to the connection nodes of adjacent repeating units within the polymer is given.

[0071] (4) Polymer graph edge attribute information assignment step

[0072] Furthermore, the polymer graph information processing method of the present invention can have a polymer edge attribute variable setting step of setting, as polymer edge attribute variables, the information of the edges connecting the nodes corresponding to each atom constituting the repeating unit that forms the selected polymer to each other and the connection edges connecting at least one of the nodes and the connection nodes. The connection edge is an edge connecting a connection node and another node.

[0073] Furthermore, the present invention has a polymer graph edge attribute information assignment step of assigning the attribute values of the edges and the connection edges to the set polymer edge attribute variables. To the connection edge, a connection edge attribute value meaning that it connects a connection node and another node is assigned as its edge attribute variable.

[0074] 3. Method for predicting the glass transition temperature of a polymer using the graph representation of the polymer according to the present invention

[0075] Next, a method for processing polymer data in an artificial neural network will be described according to the above-described polymer graph representation method of the present invention.

[0076] The present invention provides a method for constructing an artificial neural network model for predicting the glass transition temperature of a polymer by applying the above-described graph representation method of the polymer, and performing machine learning to predict the glass transition temperature of the polymer.

[0077] 3.1. Method for Generating Glass Transition Temperature Prediction Model of Polymer

[0078] Figure 4(a) shows a method for generating a glass transition temperature prediction model of a polymer according to the present invention. Based on this, a method for generating a computer-implemented glass transition temperature prediction model of a polymer for predicting the glass transition temperature of a selected polymer from its chemical structure according to the graph information representation method of the present invention will be described.

[0079] The glass transition temperature prediction model of the present invention is configured to predict the glass transition temperature by applying the relationship between the glass transition temperature and the molecular weight of the polymer defined by the Flory-Fox equation of Equation 1. The glass transition temperature T of the polymer g is a physical property value that is affected by both the molecular structure and the molecular weight. However, the conventional technology could not reflect both the influence of the molecular structure and the molecular weight when predicting the glass transition temperature. The present invention uses the following Equation 1 to reflect both the molecular structure and the molecular weight in the glass transition temperature prediction, thereby realizing a prediction model with higher accuracy. For polymers with the same molecular structure, it has become possible to predict the change in T g accompanying the change in molecular weight.

[0080]

Equation

[0081] (T g,infis the glass transition temperature when the molecular weight is theoretically infinite, K is an experimental parameter related to the free volume in the polymer, and M n is the number average molecular weight of the polymer.)

[0082] Therefore, the prediction model of the present invention includes an artificial neural network that predicts T g,inf and K in the above formula (1) from the molecular structure of the selected polymer, and the T g,inf and K values output by the artificial neural network and the molecular weight of the selected polymer are applied to the above formula (1) to calculate the glass transition temperature T g of the selected polymer. In such a prediction model, the following procedure is performed to generate an artificial neural network model that predicts T g,inf and K values, which are the glass transition temperature characteristic information of the polymer.)

[0083] (1) Preparation process of learning data (T10)

[0084] In order to generate the above-mentioned glass transition temperature prediction model of the polymer, learning data for training the artificial neural network is prepared. The learning data of the present invention consists of a data set of {polymer graph information including the structural information of the sample polymer, the measured T g value of the sample polymer, and the molecular weight M n of the sample polymer}.

[0085] Furthermore, the present invention includes, as learning data, a data set composed of the above {polymer graph information including the structural information of the sample polymer, the measured T g value of the sample polymer, and the molecular weight M n of the sample polymer}, and is characterized by including at least a predetermined number of data sets of polymers having different molecular weights while having a similar structure and data sets of polymers having the same molecular weight while having different structures. By constructing the learning data set in this way, the artificial neural network of the present invention described below can predict the T g value (T gThe predicted value) can be learned to reflect both the structural characteristics and the molecular weight characteristics of the polymer.

[0086] (2) Construction of artificial neural network and learning data input step (T20)

[0087] Next, an artificial neural network model to be trained using the learning data is constructed. As the artificial neural network in the present invention, a known artificial neural network can be used. In particular, an artificial neural network such as a message passing neural network (MPNN) or a graph neural network (GNN) based on message passing can be applied.

[0088] For example, when applying GNN as the artificial neural network, in the case of supervised learning, for the graph neural network (GNN), the polymer graph information including the structural information of the sample polymer constructed previously, the measured T g value of the sample polymer, and the molecular weight M n} of the sample polymer are input as the molecular learning data.

[0089] At this time, the polymer graph information is input as the input value of the artificial neural network, and the known measured T g value and the molecular weight M n of the polymer are input as the true values (labeled data) of the output values of the neural network. At this time, the input polymer graph information includes at least one of the polymer graph node connection information, polymer graph node attribute information, and polymer graph edge attribute information, and can be input after being embedded in the polymer adjacency matrix and polymer node / edge characteristic matrix.

[0090] (3) Learning step of artificial neural network (T30)

[0091] This is a process of updating the neural network parameters of the artificial neural network based on the input polymer graph information, molecular weight information, and glass transition temperature characteristic information of the polymer, and training the artificial neural network. The training process of the artificial neural network conforms to the known artificial neural network training process. For example, upon receiving the input of the polymer graph information of the training data, the predicted value of the glass transition temperature characteristic information of the polymer calculated from the artificial neural network reflects the measured molecular weight information of the training data set according to the above formula 1, and T g The predicted value is calculated, and T g The neural network parameters of the artificial neural network are updated so that the error function defined as the difference between the predicted value and the measured T g value of the polymer in the training data is minimized. By updating such neural network parameters, an artificial neural network for calculating glass transition temperature characteristic information from the polymer structure is completed.

[0092] Thus, in the present invention, the artificial neural network is trained so as to minimize the error between the predicted value reflecting the molecular weight information and the measured value, so that accurate T g prediction in the polymer can be performed. g

[0093] On the other hand, in the present invention, the term "neural network parameters" of the artificial neural network includes the weighted values (weights) and biases adjusted during the learning of the artificial neural network model from the training data. Such parameters are adjusted so that the neural network model predicts a more accurate output from the input data, but are used in the sense of the ordinary terms in the art indicating that they are optimized to predict the optimal output during the machine learning process.

[0094] 3.2. Method for Predicting Polymer Glass Transition Temperature ​

[0095] Input the graph information of the polymer selected in the artificial neural network generated according to the method for generating a model of the glass transition temperature of the present invention, and the number average molecular weight M of the selected polymer n is reflected, and a procedure for predicting the glass transition temperature of the selected polymer will be described.

[0096] (1) Polymer graph information preprocessing process (S10)

[0097] As shown in Fig. 4(b), a preprocessing process is performed to obtain polymer graph information from the chemical structure of the selected polymer for which the glass transition temperature is to be predicted. As described above, the polymer graph information includes, in addition to the node and edge information of the monomers constituting the polymer, the information of the connection nodes and connection edges 40 between the repeated monomers. That is, the polymer graph information is preprocessed to include at least one of the above-described polymer graph node connection information, polymer graph node attribute information, and polymer graph edge attribute information.

[0098] (2) Polymer graph information input step (S20)

[0099] This is a step of inputting the polymer graph information based on the chemical structure of the selected polymer into the neural network model learned previously. At this time, the input polymer graph information includes at least one of the above-described polymer graph node connection information, polymer graph node attribute information, and polymer graph edge attribute information, and can be input after being embedded in the above-described polymer adjacency matrix and polymer node / edge characteristic matrix.

[0100] (3) Polymer glass transition temperature characteristic information prediction step (S30)

[0101] This is a step of calculating and outputting the glass transition temperature characteristic information (T g,inf and K value) of the selected polymer from the input polymer graph information of the selected polymer by the learned artificial neural network model.

[0102] (4) Molecular weight information reflection step (S40)

[0103] From the glass transition temperature characteristic information (T g,inf and K value) calculated by the artificial neural network model, the glass transition temperature T g is calculated using the above formula (1). The molecular weight of the polymer is obtained and reflected by a known method.

[0104] 4. Polymer glass transition temperature calculation system using the graph representation of the polymer according to the present invention

[0105] (1) Data input unit 100

[0106] The data input unit 100 is a component that receives the input of polymer graph information describing the chemical structure of the selected polymer in graphic data and the molecular weight data of the polymer, and further includes a data preprocessing unit (not shown) that converts the chemical structure of the polymer into polymer graph information and inputs this into the data input unit. The molecular weight data (molecular weight information) can be obtained by a known method.

[0107] When inputting the polymer graph information into the artificial neural network model unit described later, the data preprocessing unit may generate and input a polymer adjacency matrix representing the connection relationship between two or more nodes representing each atom constituting the single molecule and the connection relationship information between the connected nodes and the nodes, a polymer edge attribute matrix including one or more edges representing the bonds between the respective nodes and the attribute information of the connection edges representing the connections between the nodes and the connected nodes.

[0108] At this time, as described above, the polymer graph information is polymer graph node connection information representing the connection relationship between nodes corresponding to each atom constituting the repeating unit that forms the polymer and connection nodes connected to connection points that are locations where the repeating unit is repeatedly connected, polymer node attribute information representing the attribute information of nodes corresponding to each atom constituting the repeating unit that forms the polymer and the attributes of connection nodes connected to connection points that are locations where the repeating unit is repeatedly connected, and polymer edge attribute information representing the attributes of edges connecting nodes corresponding to each atom constituting the repeating unit that forms the polymer to each other and connection edges connecting at least one of the nodes to the connection nodes, and may include at least one of them.

[0109] (2) Artificial Neural Network Model Unit 200

[0110] It is a component that predicts and calculates the glass transition temperature characteristic information of the polymer from the polymer graph information. The glass transition temperature characteristic information of the polymer corresponds to T and K in the above-mentioned mathematical formula 1. g,inf and K.

[0111] Such an artificial neural network model unit is configured to receive an input of polymer graph information and output a predicted value of the glass transition temperature characteristic information of the polymer, calculate a predicted value of the glass transition temperature characteristic information of a predetermined learning polymer from the polymer graph information of the learning polymer, calculate a predicted T value from the calculated predicted value of the glass transition temperature characteristic information of the learning polymer and the measured molecular weight information of the learning polymer, and update its neural network parameters so that an error function defined as the difference between the predicted T value and the measured T value of the polymer in the learning data is minimized, and is provided with a polymer glass transition temperature characteristic information prediction artificial neural network that is machine-learned. g value, and predicted T g value and the measured T of the polymer in the learning data g value, and is machine-learned by updating its neural network parameters so that an error function defined as the difference therebetween is minimized, and includes a polymer glass transition temperature characteristic information prediction artificial neural network.

[0112] (3) Glass Transition Temperature Calculation Unit 300

[0113] The glass transition temperature calculation unit is a component that calculates the glass transition temperature of a polymer according to the above-mentioned formula (1) from the polymer glass transition temperature characteristic information calculated by the artificial neural network model unit and the molecular weight data of the polymer input from the data input unit.

[0114] As described above, a method and a system for graphically representing the structure of a polymer according to the present invention, predicting the glass transition temperature characteristic information of a polymer from the graph information of the polymer, reflecting the molecular weight information of the polymer in the predicted glass transition temperature characteristic information of the polymer, and finally calculating the accurate glass transition temperature of the polymer have been described.

[0115] According to the method and system for calculating the glass transition temperature of a polymer reflecting the molecular weight information according to the present invention as described above, it is possible to separately consider the influence of the molecular weight and adjust the prediction according to the molecular weight. As a result, an effect is brought about that it is possible to construct a model that can be accurately predicted.

[0116] For example, when the actual T of samples of three polymers A, B, and C having similar structures and molecular weights of 2,000, 50,000, and 80,000 respectively g is similar, according to the model of the prior art, since only the structure of the molecule is reflected without considering the molecular weight, for samples having similar structures, the predicted T g values are learned and predicted to be similar regardless of the molecular weight.

[0117] TIFF2025519054000011.tif79170

[0118] TIFF2025519054000012.tif51170

[0119] TIFF2025519054000013.tif79170

[0120] TIFF2025519054000014.tif41170

[0121] Thus, according to the present invention, since the neural network model is learned to predict the glass transition temperature by reflecting not only the molecular structure of the polymer but also its molecular weight information, the respective influences of the molecular structure information and the molecular weight information are learned, and it becomes possible to predict a glass transition temperature that is even more accurate than the conventional technology.

Explanation of symbols

[0122] 10 Node 20 Edge 30 Connection point 40 Connection edge 100 Data input section 200 Artificial neural network model section 300 Glass transition temperature calculation section

Claims

1. A method for generating a computer-implemented model for predicting the glass transition temperature of a selected polymer, comprising: a learning data acquisition step of constructing a learning dataset having structural information of a plurality of learning polymers, known glass transition temperature measurement values of the polymers, and measured molecular weights of the polymers; a step of constructing an artificial neural network for predicting polymer glass transition temperature characteristic information, which receives an input of polymer structural information and outputs a predicted value of the glass transition temperature characteristic information of the polymer; a learning data input step of inputting the polymer structural information of the learning polymer and the measured glass transition temperature value of the learning polymer into the artificial neural network for predicting polymer glass transition temperature characteristic information; a learning step of the artificial neural network for predicting polymer glass transition temperature characteristic information, in which the neural network parameters of the artificial neural network for predicting polymer glass transition temperature characteristic information are updated based on a comparison result between a predicted glass transition temperature value calculated by reflecting the measured molecular weight of the polymer in the learning dataset among the predicted values of the glass transition temperature characteristic information of the learning polymer output by the artificial neural network for predicting polymer glass transition temperature characteristic information and the measured value of the glass transition temperature of the learning polymer; A method for generating a prediction model of the glass transition temperature of a polymer, including the above steps.

2. The glass transition temperature characteristic information is the T value and the K value in the following mathematical formula 1, and is a method for generating a glass transition temperature prediction model of the polymer according to claim 1. ｇ，ｉｎｆ ​ 【Number 10】 (T ｇ is the glass transition temperature of a polymer having a molecular weight of M ｎ , T ｇ，ｉｎｆ is the glass transition temperature when the molecular weight is infinite, K is a free volume related variable of the polymer, and M ｎ is the number average molecular weight of the polymer.)

3. The polymer structural information is: converted into polymer graph information and input into the artificial neural network for predicting polymer glass transition temperature characteristic information, The polymer graph information includes: polymer graph node connection information, which is information representing the mutual connection relationship between nodes corresponding to each atom constituting the repeating unit forming the polymer and connection nodes connected to connection points where the repeating unit is repeatedly connected; polymer node attribute information, which is information representing the attribute information of nodes corresponding to each atom constituting the repeating unit forming the polymer and the attributes of connection nodes connected to connection points where the repeating unit is repeatedly connected; Polymer edge attribute information, which is information representing the attributes of edges connecting nodes corresponding to each atom constituting the repeating unit forming the polymer and connection edges connecting at least one of the nodes and the connection nodes, The method for generating a glass transition temperature prediction model of a polymer according to claim 2, including at least one of the above.

4. A computer-implemented method for calculating a predicted glass transition temperature of a selected polymer, comprising: A step of obtaining a polymer glass transition temperature characteristic information prediction artificial neural network that generates a graph artificial neural network trained to predict the glass transition temperature characteristic information of the polymer based at least in part on a chemical structure related to the selected polymer; A step of obtaining polymer graph information that describes the chemical structure of the selected polymer by graphic data; A polymer graph input step of providing the obtained polymer graph information as an input to the graph neural network; A step of obtaining polymer glass transition temperature characteristic information of receiving the polymer glass transition temperature characteristic information of the selected polymer from the output of the graph artificial neural network; A polymer glass transition temperature calculation step of substituting the molecular weight information of the selected polymer into the polymer glass transition temperature characteristic information to calculate the glass transition temperature of the selected polymer; A computer-implemented method including the above.

5. The polymer graph information that describes the chemical structure of the selected polymer by graph information includes: Polymer graph node connection information, which is information representing the connection relationship between nodes corresponding to each atom constituting the repeating unit forming the polymer and connection nodes connected to connection points where the repeating unit is repeatedly connected; Polymer node attribute information, which is information representing the attribute information of nodes corresponding to each atom constituting the repeating unit forming the polymer and the attributes of connection nodes connected to connection points where the repeating unit is repeatedly connected; Polymer edge attribute information, which is information representing the attributes of edges connecting nodes corresponding to each atom constituting the repeating unit forming the polymer and connection edges connecting at least one of the nodes and the connection nodes; The computer-implemented method according to claim 4, comprising at least one of them.

6. The polymer graph input step of providing the obtained polymer graph information as an input to the machine-learned graph neural network includes: A polymer adjacency matrix construction step of constructing an adjacency matrix representing the connection relationship between two or more nodes representing each atom constituting the single molecule; A polymer adjacency matrix input step of inputting the constructed graph matrix as the graph neural network; including The polymer adjacency matrix The computer-implemented method according to claim 5, further including connection relationship information between the connected node and the node.

7. The graph artificial neural network calculates a predicted value of the glass transition temperature characteristic information of the learning polymer from the structural information of a plurality of learning polymers, and updates its neural network parameters so as to minimize the error between the predicted value of the glass transition temperature calculated by reflecting the measured molecular weight of the polymer in the calculated predicted value of the glass transition temperature characteristic information and the measured value of the glass transition temperature of the learning polymer. The computer-implemented method according to claim 4, which has been learned.

8. The step of obtaining the polymer glass transition temperature characteristic information prediction artificial neural network includes: a step of obtaining learning data including the chemical structures of a plurality of exemplary polymers and the measured values of the glass transition temperature and molecular weight of the exemplary polymers having the chemical structures of the exemplary polymers by one or more computing devices; a learning step of obtaining graph information graphically describing the chemical structure of the exemplary polymer, inputting it into a graph neural network, and learning the graph neural network so that a predicted value of the glass transition temperature characteristic information of the exemplary polymer output by the graph neural network is output; The computer-implemented method according to claim 4, including

9. The graph information graphically describing the chemical structure of the exemplary polymer includes: two or more node information representing each atom constituting the single molecule forming the polymer; one or more edge information representing the bonds between the respective atoms; connection node information which is a location where the single molecule is repeatedly connected; connection edge information representing the connection between the atom and the connection node The computer-implemented method according to claim 8, comprising

10. The glass transition temperature characteristic information is T in the following mathematical formula 1 ｇ，ｉｎｆ value and K, the computer-implemented method according to any one of claims 4 to 9. 【Number 11】 (T ｇ is the glass transition temperature of a polymer having a molecular weight of M ｎ , T ｇ，ｉｎｆ is the glass transition temperature when the molecular weight is infinite, K is the free volume relationship variable of the polymer, and M ｎ is the number average molecular weight of the polymer.)

11. The computer-implemented method according to any one of claims 4 to 9, wherein the polymer glass transition temperature calculation step includes calculating the glass transition temperature of the polymer by substituting the number average molecular weight information of the polymer into the following mathematical formula 1. 【Number 12】 (T ｇ is the glass transition temperature of a polymer having a molecular weight of M ｎ , T ｇ，ｉｎｆ is the glass transition temperature when the molecular weight is infinite, K is the free volume related variable of the polymer, and M ｎ is the number average molecular weight of the polymer.)

12. A data input unit that receives input of the structural information of the selected polymer and the molecular weight data of the polymer, An artificial neural network model unit that calculates a predicted value of the glass transition temperature characteristic information of the polymer from the structural information of the polymer, A glass transition temperature calculation unit that calculates a predicted value of the glass transition temperature of the polymer from the predicted value of the glass transition temperature characteristic information of the polymer and the molecular weight data of the polymer, A polymer glass transition temperature calculation system comprising the above.

13. The glass transition temperature characteristic information of the polymer is the T in the following mathematical formula 1 ｇ，ｉｎｆ value and K, and The molecular weight data of the polymer is M in the following formula (1). ｎ and The glass transition temperature calculation unit The glass transition temperature T is calculated from the following formula 1. ｇ The glass transition temperature calculation system for a polymer according to claim 12, which calculates the glass transition temperature. 【Number 13】 (T ｇ is the glass transition temperature of a polymer having a molecular weight of M ｎ , T ｇ、ｉｎｆ is the glass transition temperature when the molecular weight is infinite, K is the free volume related variable of the polymer, and M ｎ is the number average molecular weight of the polymer.)

14. The polymer glass transition temperature calculation system according to claim 13, further comprising a data preprocessing unit that converts the structural information of the polymer into polymer graph information, The artificial neural network model unit is machine-learned to calculate a predicted value of the glass transition temperature characteristic information of the polymer from the polymer graph information, The polymer graph information Polymer graph node connection information, which is information representing the connection relationship between nodes corresponding to each atom constituting the repeating unit forming the polymer and connection nodes connected to connection points where the repeating unit is repeatedly connected among the nodes, Polymer node attribute information, which is information representing the attribute information of nodes corresponding to each atom constituting the repeating unit forming the polymer and the attributes of connection nodes connected to connection points where the repeating unit is repeatedly connected among the nodes, Polymer edge attribute information, which is information representing the attributes of edges connecting nodes corresponding to each atom constituting the repeating unit forming the polymer to each other and connection edges connecting at least one of the nodes to the connection nodes, The polymer glass transition temperature calculation system according to claim 13, including at least one of the above.

15. The artificial neural network model unit Comprises a polymer glass transition temperature characteristic information prediction artificial neural network configured to receive input of polymer structure information and output a predicted value of the glass transition temperature characteristic information of the polymer. The artificial neural network for predicting polymer glass transition temperature characteristics uses the structural information of a predetermined learning polymer and the measured glass transition temperature and molecular weight of the learning polymer as learning data to calculate a predicted value of the glass transition temperature characteristics of the learning polymer from the structural information of the learning polymer, and the artificial neural network for predicting polymer glass transition temperature characteristics calculates a predicted glass transition temperature of the learning polymer using the predicted value of the glass transition temperature characteristics and the measured molecular weight, and updates the neural network parameters of the artificial neural network for predicting polymer glass transition temperature characteristics by comparing the measured glass transition temperature of the learning polymer with the calculated value, and is machine-learned. The polymer glass transition temperature calculation system according to claim 12.

16. The learning data Polymer graph information including the structural information of the sample polymer, and the measured T ｇ value of the sample polymer and the molecular weight M ｎ of the sample polymer, and includes a dataset containing the same. at least a predetermined number of the data sets are data sets of polymers having different molecular weights while having similar structures and data sets of polymers having the same molecular weight while having different structures. The polymer glass transition temperature calculation system according to claim 15.

Images