Material structure segmentation coding method capable of being used for machine learning

Abstract

The invention provides a material structure segmentation coding method for machine learning, and the method comprises the steps: analyzing a key atomic structure of a material structure, extracting a skeleton unit of the material structure, and constructing a skeleton model; dividing the skeleton unit into a plurality of fragment units, and performing unique coding according to the skeleton unit and the fragment units to obtain skeleton coding information and fragment coding information; generating a coding matrix of the material structure according to the skeleton coding information and the fragment coding information; and training the machine learning model by using the coding matrix to obtain a target machine learning model which can perform material property prediction. According to the method, multi-dimensional material information processing is realized through a mode of two-stage segmentation and datamation coding of one or more frameworks and one or more fragments, the accuracy of material property prediction and the universality of material structure coding are remarkably improved, the material design process is accelerated, and the method is suitable for material scientific research and efficient discovery of new materials.

Description

Technical Field

[0002] The present invention relates to the technical field of material science and machine learning, and in particular to a material structure segmentation encoding method that can be used for machine learning. Background Art

[0003] The purpose of material design is to achieve specific functions or meet specific needs by systematically adjusting the structure, composition and properties of materials. Traditional material design relies on experiments and theoretical simulations, which is a time-consuming and laborious process. With the rise of machine learning in materials science, researchers have accelerated the material discovery and design process by utilizing large amounts of experimental data and material databases combined with machine learning algorithms to establish a relationship model between material structure and properties. The first step in establishing such a model is to reasonably segment and encode the structure of the material so that meaningful structure-performance relationships can be extracted from it.

[0004] In terms of material segmentation and coding, existing methods mainly rely on manual processing or rule-based algorithms, which are inefficient and lack universality. In addition, coding is usually based on specific parameter descriptions such as lattice parameters and atomic coordinates, but the high dependence on specific parameter descriptions makes the model have problems such as poor adaptability and insufficient information expression.

[0005] Therefore, there is an urgent need for a material structure segmentation and coding method that can realize multi-dimensional material information processing and improve the versatility and characterization accuracy of material structure coding. Summary of the invention

[0006] Based on this, it is necessary to provide a material structure segmentation encoding method that can be used for machine learning to address the above technical problems.

[0007] A material structure segmentation coding method that can be used for machine learning includes the following steps: analyzing the key atomic structure of the material structure, extracting the skeleton unit of the material structure, and constructing a skeleton model; dividing the skeleton unit into a plurality of fragment units, and uniquely encoding the skeleton unit and the fragment unit to obtain skeleton coding information and fragment coding information; generating a coding matrix of the material structure according to the skeleton coding information and the fragment coding information; and training a machine learning model using the coding matrix to obtain a target machine learning model, wherein the target machine learning model can predict material properties.

[0008] In one embodiment, the key atomic structure of the material structure is analyzed, the skeleton units of the material structure are extracted, and the skeleton model is constructed, including: measuring and analyzing the spatial arrangement of atoms in the material structure, and finding the key atomic structure that forms the skeleton; forming the skeleton units based on the key atomic structure, and constructing the skeleton model of the material according to the skeleton units.

[0009] In one of the embodiments, the skeleton unit is divided into a number of fragment units, and the skeleton units and fragment units are uniquely encoded to obtain skeleton coding information and fragment coding information, including: analyzing the internal structure of the skeleton unit, dividing the key atomic structure within the skeleton unit into a number of fragment units; uniquely encoding the skeleton units and fragment units respectively to obtain skeleton coding information and fragment coding information, wherein the skeleton coding information and fragment coding information correspond to different skeleton types and fragment types respectively.

[0010] In one of the embodiments, generating a coding matrix of the material structure based on the skeleton coding information and the fragment coding information includes: using the skeleton coding information as row elements and the fragment coding information as column elements to generate a coding matrix representing the material structure, wherein the coding matrix has the same dimension as the material structure.

[0011] In one embodiment, the encoding matrix is ​​used to train the machine learning model to obtain a target machine learning model, including: using the encoding matrix as input to train the machine learning model; optimizing model parameters through error back propagation until the model error reaches a set threshold, and verifying the model through test set data to obtain a target machine learning model, wherein the target machine learning model is used to predict material properties, and the material properties include physical properties and chemical properties.

[0012] Compared with the prior art, the advantages and beneficial effects of the present invention are as follows: by analyzing the key atomic structure of the material structure, extracting the skeleton unit of the material structure, constructing a skeleton model, dividing the skeleton unit into a number of fragment units, uniquely encoding the skeleton unit and the fragment unit to obtain skeleton coding information and fragment coding information, generating a coding matrix of the structure according to the skeleton coding information and the fragment coding information, and through the two-level segmentation and data coding of (one or more) skeletons-(one or more) fragments, it is possible to more efficiently characterize complex material structures, which is suitable for design scenarios of various materials and easy to integrate with machine learning models. The machine learning model is trained using the coding matrix to obtain a target machine learning model, and material properties can be predicted through the target machine learning model. Through multi-level skeleton and fragment coding, multi-dimensional material information processing can be performed, which simplifies the preprocessing process of model input data, significantly improves the accuracy of material structure characterization and material design efficiency, improves the versatility of material structure coding, and is suitable for various material design and optimization scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 A schematic diagram of a flow chart of a material structure segmentation encoding method that can be used for machine learning in one embodiment; Figure 2A schematic diagram of a process flow of a skeleton-fragment two-level segmentation mode in one embodiment; Figure 3 A schematic diagram of two types of atoms and key skeletons in a graphene-like structure in one embodiment; Figure 4 A schematic diagram of the combination of fragments in a triangular unit skeleton in one embodiment; DETAILED DESCRIPTION

[0014] Before describing the specific embodiments of the present invention, the overall concept of the present invention is described as follows: The present invention is mainly developed for material segmentation and coding processes, which currently rely mainly on manual processing or rule-based algorithms, are inefficient and lack versatility.

[0015] Therefore, the present invention proposes a material structure segmentation and coding method that can be used for machine learning. The key atomic structure of the material structure is analyzed, the skeleton unit of the material structure is extracted, and a skeleton model is constructed. The skeleton unit is divided into a number of fragment units, and unique encoding is performed according to the skeleton unit and the fragment unit to obtain skeleton coding information and fragment coding information. According to the skeleton coding information and the fragment coding information, a coding matrix of the structure is generated. Through the skeleton-fragment two-level segmentation and data coding, complex material structures can be more efficiently characterized, which is suitable for design scenarios of various materials and is easy to integrate with machine learning models. The machine learning model is trained using the coding matrix to obtain a target machine learning model. Material properties can be predicted through the target machine learning model. Multi-level skeleton and fragment coding can be used to perform multi-dimensional material information processing, which simplifies the preprocessing process of model input data, significantly improves the accuracy of material structure characterization and material design efficiency, and improves the versatility of material structure coding, which is suitable for various material design and optimization scenarios.

[0016] After introducing the overall concept of the present invention, in order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below through specific implementation methods combined with the accompanying drawings. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0017] In one embodiment, Figure 1 As shown, a material structure segmentation encoding method that can be used for machine learning is provided, comprising the following steps: Step S110, analyzing the key atomic structure of the material structure, extracting the skeleton units of the material structure, and constructing a skeleton model.

[0018] Specifically, in order to achieve efficient characterization of the material structure, it is necessary to analyze the key atomic structure of the material structure, extract the skeleton units that form the material skeleton, determine the key atomic structure that forms the skeleton by calculating spatial parameters such as atomic spacing and bond angles, and define it as a skeleton unit. The skeleton model is constructed through the skeleton units to facilitate subsequent skeleton encoding.

[0019] Among them, step S110 includes: measuring and analyzing the spatial arrangement of atoms in the material structure, finding the key atomic structure that forms the skeleton; forming skeleton units based on the key atomic structure, and constructing a skeleton model of the material based on the skeleton units.

[0020] Specifically, by measuring and analyzing the spatial arrangement of atoms in the material lattice, the key atomic structure that forms the skeleton is found. The key atomic structure is the main supporting structure of the material. The skeleton unit is extracted based on the key atomic structure, and a skeleton model of the entire material structure is constructed. In actual use, the initial material model can be generated using software (such as crystal analysis software VESTA or CrystalMaker), and the skeleton model can be extracted through parameter setting.

[0021] Step S120, dividing the skeleton unit into a plurality of fragment units, performing unique coding according to the skeleton unit and the fragment unit, and obtaining skeleton coding information and fragment coding information.

[0022] Specifically, Figure 2 As shown, the skeleton unit is divided into several fragment units, each fragment unit represents a unique substructure or component distribution, and the skeleton unit and the fragment unit are uniquely encoded respectively to obtain skeleton encoding information and fragment encoding information, forming a digitized encoding structure.

[0023] Among them, step S120 includes: analyzing the internal structure of the skeleton unit, dividing the key atomic structure within the skeleton unit into several fragment units; uniquely encoding the skeleton unit and the fragment unit respectively to obtain skeleton coding information and fragment coding information, and the skeleton coding information and the fragment coding information correspond to different skeleton types and fragment types respectively.

[0024] Specifically, the internal structure of the skeleton unit is analyzed, and the key atomic structure within the skeleton unit is divided into several fragment units. The skeleton units and fragment units are uniquely encoded respectively. Different skeleton types and fragment types are represented by skeleton coding information and fragment coding information respectively. For example, the fragment coding values ​​range from 0 to N to represent different fragment types, and the encoded data is stored in a matrix. The above encoding method can facilitate data input and improve model training efficiency.

[0025] like Figure 3As shown in the figure, two types of carbon atoms and key skeletons in graphene are obtained by analyzing the key atomic structure of the material structure; and further analysis of the skeleton unit can be obtained as follows Figure 4 The fragment assembly in the triangular unit skeleton is shown.

[0026] Step S130, generating a coding matrix of the material structure according to the skeleton coding information and the fragment coding information.

[0027] Specifically, a complete material structure coding matrix is ​​generated based on the skeleton coding information and fragment coding information. Each element in the coding matrix represents a unit in the material structure. Through the two-level segmentation and data coding of (one or more) skeletons-(one or more) fragments, it is possible to more efficiently characterize complex material structures and facilitate integration with machine learning models.

[0028] Wherein, step S130 includes: using the skeleton coding information as row elements and the fragment coding information as column elements to generate a coding matrix representing the material structure, and the coding matrix has the same dimension as the material structure.

[0029] Specifically, the skeleton coding information is used as the row element and the fragment coding information is used as the column element to generate a two-dimensional coding matrix representing the material structure. Here is an example, for example, the size of the coding matrix is ​​set to (2,4), and the element value is between 0 and 6, representing each coding unit type. The coding matrix can be directly input into the machine learning model to predict material properties.

[0030] Step S140, use the encoding matrix to train the machine learning model to obtain a target machine learning model.

[0031] Specifically, the coding matrix is ​​used as the input of the machine learning model for model training to obtain a target machine learning model. The target machine learning model can predict material properties. By inputting the encoded data into the target machine learning model, the model is used to predict the physical or chemical properties of the material. Through multi-level skeleton and fragment encoding methods, the complex material structure is digitized to facilitate machine learning model processing, which significantly improves the accuracy of material structure characterization and material design efficiency.

[0032] Among them, step S140 includes: using the coding matrix as input to train the machine learning model; optimizing the model parameters through error back propagation until the model error reaches a set threshold, and verifying the model through test set data, and obtaining the target machine learning model through verification. The target machine learning model is used to predict material properties, and the material properties include physical properties and chemical properties.

[0033] Specifically, any suitable machine learning algorithm is selected, such as a neural network model, and the encoding matrix is ​​input into the machine learning model for model training. The model parameters are adjusted through multiple rounds of training, and the model parameters are optimized through error back propagation until the model error reaches the set threshold, thereby obtaining a model that meets the prediction accuracy. After the model training is completed, the test set data is input to verify the model, and the model performance is evaluated by the test set error. After the verification, the target machine learning model is obtained, and the material properties can be predicted through the target machine learning model, including but not limited to physical properties and chemical properties, which significantly improves the accuracy of material structure characterization and material design efficiency.

[0034] In this embodiment, the key atomic structure of the material structure is analyzed, the skeleton unit of the material structure is extracted, and a skeleton model is constructed. The skeleton unit is divided into several fragment units, and unique encoding is performed according to the skeleton unit and the fragment unit to obtain skeleton coding information and fragment coding information. According to the skeleton coding information and the fragment coding information, a coding matrix of the structure is generated. Through the skeleton-fragment two-level segmentation and data coding, complex material structures can be more efficiently characterized, which is suitable for design scenarios of various materials and is easy to integrate with machine learning models. The machine learning model is trained using the coding matrix to obtain a target machine learning model. Material properties can be predicted through the target machine learning model. Through multi-level skeleton and fragment coding, multi-dimensional material information processing can be performed, which simplifies the preprocessing process of model input data, significantly improves the accuracy of material structure characterization and material design efficiency, improves the versatility of material structure coding, and is suitable for various material design and optimization scenarios.

[0035] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

[0036] Obviously, those skilled in the art should understand that the modules or steps of the present invention described above can be implemented by a general-purpose computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, and optionally, they can be implemented by a program code executable by a computing device, so that they can be stored in a computer storage medium (ROM / RAM, magnetic disk, optical disk) and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order from that herein, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Therefore, the present invention is not limited to any specific combination of hardware and software.

[0037] The above contents are further detailed descriptions of the present invention in combination with specific implementation methods, and it cannot be determined that the specific implementation of the present invention is limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, several simple deductions or substitutions can be made without departing from the concept of the present invention, which should be regarded as falling within the scope of protection of the present invention.

Claims

1. A material structure segmentation encoding method that can be used for machine learning, characterized in that: The following steps are involved: Analyze the key atomic structure of the material structure, extract the skeleton units of the material structure, and construct a skeleton model; Divide the skeleton unit into a plurality of fragment units, and perform unique coding according to the skeleton unit and the fragment unit to obtain skeleton coding information and fragment coding information; Generate a coding matrix of the material structure according to the skeleton coding information and the fragment coding information; The coding matrix is ​​used to train the machine learning model to obtain a target machine learning model, and the target machine learning model can predict material properties.

2. A material structure segmentation coding method that can be used for machine learning according to claim 1, characterized in that: The key atomic structure of the material structure is analyzed, the skeleton unit of the material structure is extracted, and the skeleton model is constructed, including: Measure and analyze the spatial arrangement of atoms in the material structure to find the key atomic structure that forms the skeleton; A skeleton unit is formed based on the key atomic structure, and a skeleton model of the material is constructed according to the skeleton unit.

3. A material structure segmentation coding method that can be used for machine learning according to claim 2, characterized in that: The step of dividing the skeleton unit into a plurality of fragment units and performing unique coding according to the skeleton unit and the fragment unit to obtain skeleton coding information and fragment coding information includes: Analyzing the internal structure of the skeleton unit, and dividing the key atomic structure in the skeleton unit into a number of fragment units; The skeleton unit and the fragment unit are respectively uniquely encoded to obtain skeleton encoding information and fragment encoding information, wherein the skeleton encoding information and the fragment encoding information correspond to different skeleton types and fragment types respectively.

4. The material structure segmentation coding method that can be used for machine learning according to claim 1 is characterized in that: The step of generating a coding matrix of the material structure according to the skeleton coding information and the fragment coding information comprises: The skeleton coding information is used as row elements and the fragment coding information is used as column elements to generate a coding matrix representing the material structure, and the coding matrix has the same dimension as the material structure.

5. The material structure segmentation coding method that can be used for machine learning according to claim 1 is characterized in that: The method of training the machine learning model using the encoding matrix to obtain a target machine learning model includes: Using the encoding matrix as input, training a machine learning model; The model parameters are optimized through error back propagation until the model error reaches the set threshold. The model is verified through test set data. The target machine learning model is obtained through verification. The target machine learning model is used to predict material properties, including physical properties and chemical properties.

Images