Crystal structure optimization application method based on graph convolutional neural network generation model

Abstract

The invention discloses a crystal structure optimization application method based on a graph convolutional neural network generation model, and relates to the technical field of crystal structures, and the method comprises the steps: obtaining initial crystal structure data, carrying out the data augmentation of the initial crystal structure data, and obtaining the feature data of each to-be-screened crystal structure; training the graph convolutional neural network, performing primary screening on the feature data of each to-be-screened crystal structure according to the graph convolutional neural network to obtain each target crystal structure, and obtaining feature values of each target crystal structure; and performing stability verification on each target crystal structure to obtain a performance evaluation value of each target crystal structure, performing secondary screening according to each target crystal structure characteristic value and each target crystal structure performance evaluation value to obtain each qualified crystal structure, and performing feedback output. According to the invention, through an intelligent mode based on data driving, the efficiency and accuracy of crystal structure optimization are greatly improved, and the consumption of manpower and material resources is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of crystal structure prediction, and specifically to an application method for crystal structure optimization based on a graph convolutional neural network generation model. Background Art

[0002] With the continuous advancement of modern science and technology and the increasingly stringent requirements of various industries on the performance of new materials, crystal structure optimization has become a key link in improving the comprehensive performance of materials. The crystal structure optimization method based on the graph convolutional neural network generation model is a highly innovative and promising technical path, but this method faces many difficulties and challenges when processing complex crystal structure data, such as high model complexity, high consumption of computing resources, and strong dependence on large-scale data.

[0003] Existing methods for crystal structure optimization mainly rely on traditional experimental tests and numerical calculation methods based on physical models. Through repeated experiments, fine-tuning of parameters, and simulation using classical computing theories, efforts are made to optimize and improve the crystal structure, thereby meeting the requirements for material performance in different application scenarios.

[0004] For example, the invention patent with announcement number CN108268750B announces a method for predicting hypothetical inorganic crystal structures based on enumerating Wyckoff position combinations. The first step is to select the target space group for structure prediction and determine the range of unit cell parameters and skeleton density; the second step is to enumerate the Wyckoff position combinations; the third step is to build a structural model; under a specific space group, through the enumerated Wyckoff position combinations, the atoms are fixed at the Wyckoff positions to build a hypothetical inorganic crystal structure model, and finally the hypothetical inorganic crystal structure prediction is completed.

[0005] For example, the invention patent with publication number CN116343962A discloses a self-supporting crystal structure design optimization method based on topology optimization. The technology of the present invention mainly includes two stages: subdivision and simplification. In the subdivision stage, the method subdivides the initial rough crystal structure composed of several self-supporting crystal units based on the topology optimization framework, and determines whether the structural unit is subdivided by assigning a state variable to each crystal unit. The state variables of the units in each subdivision level are converted into the virtual density of the rod, so that they can be included in the continuous calculation. The self-support of the rod is guaranteed by the carefully designed unit topology, while the self-support on the node is achieved by the self-support filter introduced on the rod density under each newly introduced node. By repeatedly applying subdivision, an adaptive and optimized self-supporting crystal structure can be obtained.

[0006] However, in the process of implementing the technical solution of the invention in the embodiments of the present application, it was found that the above technology has at least the following technical problems: in practical applications, since the method based on enumerating Wyckoff position combinations relies on fixed rules, it may have poor adaptability to complex systems, resulting in reduced prediction accuracy and affecting the subsequent crystal research and development process. Summary of the invention

[0007] In view of the shortcomings of the prior art, the present invention provides a crystal structure optimization application method based on a graph convolutional neural network generation model, which can effectively solve the problems involved in the above-mentioned background technology.

[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions: The first aspect of the present invention provides a crystal structure optimization application method based on a graph convolutional neural network generation model, including: obtaining initial crystal structure data, and performing data augmentation on the initial crystal structure data to obtain characteristic data of each crystal structure to be screened.

[0009] The graph convolutional neural network is trained, and the characteristic data of each crystal structure to be screened are screened once according to the graph convolutional neural network to obtain each target crystal structure and obtain the characteristic value of each target crystal structure.

[0010] The stability of each target crystal structure is verified to obtain a performance evaluation value of each target crystal structure. Secondary screening is performed based on the characteristic values ​​of each target crystal structure and the performance evaluation value of each target crystal structure to obtain each qualified crystal structure and output feedback.

[0011] As a further method, the initial crystal structure data is augmented to obtain characteristic data of each crystal structure to be screened. The specific analysis process is: extracting molecular coordinate data from the initial crystal structure data, obtaining random number variables by random sampling, and transforming the molecular coordinate data according to the random number variables to obtain characteristic data of each crystal structure to be screened. The characteristic data of each crystal structure to be screened includes the molecular coordinates of each crystal structure to be screened.

[0012] As a further method, the training graph convolutional neural network has the following specific analysis process: setting up the graph convolutional neural network and initializing the graph structure; setting up the generative adversarial network, including initializing the model weights of the generator and the discriminator, initializing the loss function and optimization parameters, and setting the initialization training parameters of the generative adversarial network; importing the characteristic data of each crystal structure to be screened into the graph convolutional neural network, and training the graph convolutional neural network.

[0013] As a further method, the characteristic data of each crystal structure to be screened are screened according to the graph convolutional neural network, and the specific analysis process is: obtaining the number of critical position transformation molecules and the critical molecular transformation displacement from the materials science database; extracting the corresponding molecular coordinates from the initial crystal structure data and the characteristic data of each crystal structure to be screened, respectively, and further obtaining the transformation displacement of each molecule in each crystal structure to be screened, and comprehensively analyzing to obtain the characteristic value of each crystal structure to be screened; comparing the characteristic value of each crystal structure to be screened with the crystal structure characteristic threshold preset in the materials science database to obtain each target crystal structure.

[0014] As a further method, the characteristic value of each crystal structure to be screened is compared with the crystal structure characteristic threshold preset in the materials science database to obtain each target crystal structure. The specific analysis process is: if the characteristic value of a crystal structure to be screened is greater than or equal to the preset crystal structure characteristic threshold, the crystal structure to be screened is marked as a target crystal structure; if the characteristic value of a crystal structure to be screened is less than the preset crystal structure characteristic threshold, the crystal structure to be screened is marked as an unqualified target crystal structure; and the target crystal structures are obtained by statistics.

[0015] As a further method, the stability of each target crystal structure is verified to obtain a performance evaluation value of each target crystal structure. The specific analysis process is: the stability verification includes mechanical stability verification, thermodynamic stability verification and kinetic stability verification; the stability data of each target crystal structure is obtained through stability verification, and the stability data of each target crystal structure includes the number of negative elements of the elastic constant matrix of each target crystal structure, the energy of the crystal system at each monitoring time point, and the number of imaginary frequencies of the phonon spectrum; based on the stability data of each target crystal structure, a comprehensive analysis is performed to obtain the performance evaluation value of each target crystal structure.

[0016] As a further method, the performance evaluation value of each target crystal structure is obtained by comprehensive analysis based on the stability data of each target crystal structure. The specific analysis process is: obtaining the number of negative elements of the critical elastic constant matrix, the crystal system energy at the critical monitoring time point and the number of critical phonon spectrum imaginary frequencies from the materials science database; and obtaining the performance evaluation value of each target crystal structure based on processing the stability data of each target crystal structure. The target crystal structure performance evaluation value is used to quantitatively evaluate the stability of each target crystal structure.

[0017] As a further method, the secondary screening is performed according to the characteristic values ​​of each target crystal structure and the performance evaluation value of each target crystal structure. The specific analysis process is: according to the characteristic values ​​of each crystal structure to be screened and the performance evaluation value of each target crystal structure, a comprehensive analysis is performed to obtain the quality evaluation index of each target crystal structure; If a target crystal structure quality assessment index is greater than or equal to a preset target crystal structure quality assessment index threshold, the target crystal structure is marked as a qualified crystal structure; if a target crystal structure quality assessment index is less than a preset target crystal structure quality assessment index threshold, the target crystal structure is marked as an unqualified crystal structure; statistics of each qualified crystal structure are obtained and feedback output is performed.

[0018] As a further method, the statistics obtain each qualified crystal structure and output it as feedback. The specific process is: according to the quality evaluation index of each qualified crystal structure, the qualified crystal structures are sorted in order from large to small to obtain the output order of the qualified crystal structures; according to the output order of the qualified crystal structures, each qualified crystal structure is displayed and output.

[0019] Compared with the prior art, the embodiments of the present invention have at least the following beneficial effects: (1) The present invention provides an application method for crystal structure optimization based on a graph convolutional neural network generation model, which deeply explores the complex topological relationship between crystal atoms and transforms crystal structure optimization from a traditional rule-based and experience-based model to a data-driven intelligent model, greatly improving the efficiency and accuracy of crystal structure optimization and reducing the consumption of manpower and material resources.

[0020] (2) The present invention uses graph convolutional neural networks to model and analyze the atomic interactions and lattice arrangements in crystal structures, taking into account the complexity and diversity of crystal structures, making the optimization of crystal structures more comprehensive and in-depth, and being able to accurately predict the stability and performance of crystal structures, providing a scientific basis for the selection and modification of materials, and promoting the improvement of new material performance and the reduction of costs.

[0021] (3) The present invention establishes a dynamic feedback mechanism for crystal structure optimization to monitor and evaluate in real time the performance changes of the crystal structure during the optimization process, thereby realizing intelligent control and adaptive adjustment of the crystal structure optimization process, reducing the interference of human factors on the optimization results, and improving the stability and reliability of the optimization process, thus providing an efficient and stable solution for the industrial large-scale production of high-performance crystal materials.

[0022] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a schematic diagram of the method flow of the present invention.

[0024] Figure 2 It is a schematic diagram of the functional relationship between the target crystal structure characteristic value and the target crystal structure quality evaluation index of the present invention. DETAILED DESCRIPTION

[0025] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0026] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0027] Reference Figure 1 As shown, the first aspect of the present invention provides a crystal structure optimization application method based on a graph convolutional neural network generation model, including: obtaining initial crystal structure data, and performing data augmentation on the initial crystal structure data to obtain characteristic data of each crystal structure to be screened.

[0028] Specifically, data augmentation is performed on the initial crystal structure data to obtain characteristic data of each crystal structure to be screened. The specific analysis process is: molecular coordinate data is extracted from the initial crystal structure data, random number variables are obtained by random sampling, and the molecular coordinate data are transformed according to the random number variables to obtain characteristic data of each crystal structure to be screened, wherein the characteristic data of each crystal structure to be screened includes the molecular coordinates of each crystal structure to be screened.

[0029] In this embodiment, known crystallographic structure data is obtained from a materials science database, and the crystallographic structure is an approximate structure to the predicted crystallographic structure data; data preprocessing is performed, and before the crystallographic structure prediction is performed, all input variables are made independent of each other to improve the validity of data prediction, and data preprocessing also includes converting the three-dimensional crystallographic vectors in the original data into six crystallographic constants that are independent of each other. Randomly extracted random number variables are added on the basis of the original molecular coordinate data to achieve the purpose of translating the crystal structure in three-dimensional space. The data augmentation method includes translation transformation and rotation transformation. By adopting the method of data augmentation, the problem of serious shortage of materials science data is solved, and the crystal structure data operation is performed by directly using crystal vectors and molecular coordinates to avoid the problem of information loss caused by data dimensionality reduction in the data characterization link; for example, when performing a rotation transformation, the three molecular coordinates are mutually converted; through data augmentation, the diversity and complexity of the data set can be increased, thereby improving the generalization ability and robustness of the machine learning model.

[0030] It should be understood that in the present embodiment, the approximate structure refers to a crystal structure that is similar or close in structure, and may have certain similarities in terms of atomic arrangement, chemical bond formation or unit cell parameters, but is not completely the same. Mutually independent means that there is no direct dependency between variables, that is, the change of one variable will not affect the value of another variable. The crystal structure is composed of a large number of molecules, and the position of the molecule in the crystal is described by the molecular coordinates, which determine the position of the molecule in three-dimensional space, and the properties of the crystal are often closely related to the relative position of the molecule. For example, the distance between molecules can affect the intermolecular forces, thereby affecting the physical and chemical properties of the crystal. By extracting molecular coordinate data from the initial crystal structure data. By randomly extracting a series of random number variables, the random number variables will be used for subsequent coordinate transformations to simulate small changes in the crystal structure under different conditions. For example, different parameters can be provided for different molecular coordinate transformation operations by generating uniformly distributed or normally distributed random numbers. The random number variable can be one or more, depending on the transformation method adopted. According to the generated random number variable, the molecular coordinate data is transformed, wherein the transformation can be a translation transformation, a rotation transformation, a scaling transformation, etc. By transforming the molecular coordinate data, different characteristic data of the crystal structure to be screened are obtained, and each characteristic data of the crystal structure to be screened includes each molecular coordinate of each crystal structure to be screened.

[0031] The graph convolutional neural network is trained, and the characteristic data of each crystal structure to be screened are screened once according to the graph convolutional neural network to obtain each target crystal structure and obtain the characteristic value of each target crystal structure.

[0032] Specifically, the graph convolutional neural network is trained, and the specific analysis process is as follows: setting up the graph convolutional neural network and initializing the graph structure; setting up the generative adversarial network, including initializing the model weights of the generator and the discriminator, initializing the loss function and optimization parameters, and setting the initialization training parameters of the generative adversarial network; importing the characteristic data of each crystal structure to be screened into the graph convolutional neural network, and training the graph convolutional neural network.

[0033] In this embodiment, during the training of the graph convolutional neural network, the characteristic data of the crystal structure to be screened, the initialized graph structure, the initial values ​​of the model weights of the generator and the discriminator, the initial values ​​of the loss function and the optimization parameters, and the initialization training parameters of the generative adversarial network are input, and the trained graph convolutional neural network and the trained generative adversarial network are output. Among them, the characteristic data of the crystal structure to be screened is the basis for training the graph convolutional neural network, and contains various characteristic information of the crystal structure, such as the coordinates of atoms, the type and strength of chemical bonds, etc.; the initialized graph structure is used to construct the topological structure of the graph convolutional neural network, which describes the connection relationship between atoms in the crystal; the initial values ​​of the model weights of the generator and the discriminator need to be initialized when setting up the generative adversarial network; the initial values ​​of the loss function and the optimization parameters are used to guide the training process of the generative adversarial network, so that the network can be optimized according to the loss function; the initialization training parameters of the generative adversarial network include the learning rate, the number of iterations, and the batch size; the trained graph convolutional neural network is used to screen the characteristic data of the crystal structure; the trained generative adversarial network helps to improve the performance of the graph convolutional neural network, such as generating more diverse samples or assisting in feature extraction and discrimination. Set up a graph convolutional neural network and determine the graph structure. According to the characteristics of the crystal structure, the atoms in the crystal are used as nodes and the chemical bonds between atoms are used as edges to construct the graph structure. The adjacency matrix or edge list can be used to represent the connection relationship of the graph. For node features, various properties of atoms (such as atomic number, electronegativity, etc.) can be used as feature vectors of the nodes. Select a graph convolutional layer. Common graph convolutional layers include GCN (Graph Convolutional Network) layer and GAT (Graph Attention Network) layer. Build a network architecture, stack multiple graph convolutional layers together to form a deep network, add pooling layers (such as GraphPooling) to reduce the scale of the graph, extract higher-level features, and add fully connected layers to map the features of the graph to the required output dimensions, such as for classification or regression tasks. Select the activation function and optimizer. The activation function can be ReLU (Rectified Linear Unit), Sigmoid (S-shaped curve function) and Tanh (HyperbolicTangent), etc., and choose the appropriate activation function according to the specific task; the optimizer can use Adam (Adaptive Moment Estimation) and SGD (Stochastic Gradient Descent), etc., and adjust parameters such as the learning rate according to the training situation of the network.Set up the Generative Adversarial Network, initialize the generator and discriminator. The generator is a neural network whose input is random noise and output is generated data. Its structure can be a variant of a multi-layer perceptron (MLP) or a convolutional neural network (CNN). The discriminator is also a neural network whose input is real data or data generated by the generator. Its output is a probability value, indicating the probability that the input data is real data. Its structure can be similar to that of the generator. Initialize the model weights. You can use a random initialization method, such as normal distribution or uniform distribution, to initialize the weights of the generator and discriminator. Initialize the loss function. The commonly used loss function is the binary cross entropy loss, which is used to measure the ability of the discriminator to distinguish between real data and generated data. For the generator, its goal is to minimize the probability that the generated data is misclassified as false by the discriminator. For the discriminator, its goal is to correctly distinguish between real data and generated data. Set the optimization parameters, select a suitable optimizer, such as the Adam optimizer, and set parameters such as the learning rate and weight decay. For the generator and the discriminator, different learning rates or optimization strategies are required to balance their training processes. Set the training parameters, determine the number of training iterations, prevent overfitting or underfitting, and select the appropriate batch size, which is generally adjusted according to the size of the data set and hardware resources. The trained graph convolutional neural network can be used to screen the crystal structure feature data. The data to be screened is input into the trained network, and the network will screen according to the learned features and patterns to obtain the target crystal structure. Generative adversarial networks can assist in the training of graph convolutional neural networks. For example, samples generated by generative adversarial networks can be used as additional data to expand the training set, or help graph convolutional neural networks learn more discriminative features, thereby improving the accuracy and performance of screening. By obtaining the target crystal structure, the characteristic values ​​of the target crystal structure can be further calculated, providing a basis for subsequent research on crystal material performance analysis, structural optimization, etc.

[0034] It should be understood that the graph convolutional neural network in this embodiment is a deep learning model specially designed for processing graph structure data, and the crystal structure is adapted to the graph structure data representation because the atoms or molecules in the crystal can be regarded as nodes in the graph, and the chemical bonds or interactions between them are edges. Build GCNN (Graph Convolutional Neural Network) to process graph data, such as social networks, molecular structures, and knowledge graphs, and determine the number of network layers, the number and size of convolution kernels in each layer, the type of activation function and other hyperparameters. For example, for a social network dataset, the task is to classify nodes. Build a GCNN with 3 network layers, 32 convolution kernels of size 3 in the first layer, 64 convolution kernels of size 3 in the second layer, and 128 convolution kernels of size 3 in the third layer (or select a global average pooling layer to aggregate node features based on task requirements), and the activation function type is ReLU. This configuration can be adjusted according to the specific task and dataset. If the dataset is large and the network structure is complex, the number of network layers or convolution kernels can be increased; if the dataset is small and the task is relatively simple, the number of network layers or convolution kernels can be reduced to reduce the risk of overfitting. Initialization is to assign initial values ​​to the nodes and edges in the graph structure. For example, the nodes are assigned initial feature vectors representing information such as atomic types and charges, and the edges are set to represent the initial weights of chemical bond strength and type, providing a starting state for subsequent calculations so that the network can gradually update these values ​​to fit the data pattern. Initializing the graph structure means defining the initial features of the nodes and edges in the graph, which may include the type of atoms, the type of bonds, the spatial position, etc. The generative adversarial network consists of two core components: the generator and the discriminator. The generator is used to try to generate realistic data that looks similar to real data; the discriminator is used to try to distinguish between real data and data generated by the generator. Initializing the model weights is to assign random initial values ​​to the parameters of the neural network layers in the generator and the discriminator, such as putting starting weights on the trays on both sides of the scale. Subsequent training will continuously adjust these weights to balance them. Reasonable initial values ​​can speed up convergence, avoid falling into local optimal solutions, and allow the model to learn data features faster. The loss function usually consists of two parts: one part is used to measure the difference between the data generated by the generator and the real data, and the other part is used to measure the ability of the discriminator to distinguish between real and generated data. Common ones include mean square error and cross entropy loss. Optimization parameters involve learning rate and momentum, etc., which control the step size and direction of each parameter update and adjust the rhythm of model learning. Setting the initialization training parameters of the generative adversarial network includes the number of training rounds and batch size. The number of rounds determines the number of times the entire training process is cycled. The batch size refers to the amount of data input into the model each time. Through experiments, adjusting parameters and observing training results, we can gradually find the optimal parameter combination.For example, in the GAN training parameter setting on the MNIST dataset, the network structure is set. The generator adopts a multi-layer perceptron (MLP) or convolutional neural network (CNN) structure, the input is a random noise vector, and the output is a 28x28 pixel grayscale image. The discriminator adopts an MLP or CNN structure, the input is a 28x28 pixel grayscale image, and the output is a scalar, which represents the probability that the image is a real image. The learning rate is set to 0.0002 for the generator, which is a hyperparameter that controls the update speed of the generator parameters. A smaller learning rate means that the parameters are updated more slowly, which helps to stabilize the training process. The discriminator learning rate is 0.0002, which is the same learning rate as the generator to maintain a balance between the two. The batch size is set to 64, which is the number of samples input into the network each time training. A larger batch size can speed up the training process, but it may also increase memory consumption and computing time. The number of training rounds is set to 10,000 for the total number of training rounds, which is the number of iterations of the entire training process. More training rounds may help improve the performance of the model, but may also lead to overfitting. The discriminator is updated once every five times, which helps maintain the advantage of the discriminator and guide the generator to generate more realistic images. The noise vector dimension is set to 100, which is the dimension of the random noise vector input to the generator. A larger dimension may increase the expressive power of the generator, but may also increase the computational complexity. The optimizer is set to Adam optimizer, which is a commonly used adaptive learning rate optimization algorithm with faster convergence speed and better performance. The characteristic data of each crystal structure to be screened are imported into the graph convolutional neural network and the graph convolutional neural network is trained. The characteristic data of each crystal structure to be screened obtained in the previous processing are sent to GCNN in sequence according to the set batch size. Within the network, data is continuously propagated and the features of nodes and edges are updated according to the graph convolution operation rules. At the same time, combined with the GAN (Generative Adversarial Network) mechanism, the discriminator distinguishes between real data and data generated by the generator. The generator optimizes its own output based on the discriminator feedback, prompting GCNN to gradually fit the data distribution and achieve the purpose of selecting the target crystal structure from the data to be screened.

[0035] Specifically, the characteristic data of each crystal structure to be screened are screened according to the graph convolutional neural network. The specific analysis process is as follows: the number of critical position transformation molecules and the critical molecular transformation displacement are obtained from the materials science database; the corresponding molecular coordinates are extracted from the initial crystal structure data and the characteristic data of each crystal structure to be screened, and the transformation displacement of each molecule in each crystal structure to be screened is further obtained, and the characteristic value of each crystal structure to be screened is obtained by comprehensive analysis; the characteristic value of each crystal structure to be screened is compared with the crystal structure characteristic threshold preset in the materials science database to obtain each target crystal structure.

[0036] It should be understood that in this embodiment, the number of molecules in critical position transformation refers to the number of molecules whose positions in the crystal change when a certain critical state (such as phase change, chemical reaction, etc.) is reached during the crystal structure change process. For example, at the phase change point of the crystal, the position of some molecules changes, and the number of these molecules is the number of molecules in critical position transformation. The critical molecular transformation displacement refers to the distance of molecular position change when the crystal structure change reaches a critical state, that is, the displacement of the molecule from the original position to the new position. This displacement is critical for studying the change of the crystal structure. First, the corresponding molecular coordinates are extracted from the initial crystal structure data and the characteristic data of the crystal structure to be screened. The initial crystal structure data is used as the original reference, and the characteristic data of the crystal structure to be screened is a plurality of groups of samples after data augmentation and other processing. Based on the two sets of coordinate data, the molecular transformation displacement of each molecule in each crystal structure to be screened is obtained. The number of position transformation molecules and the molecular transformation displacement can be simulated by molecular dynamics simulation software such as LAMMPS (Large-scale Atomic / Molecular MassivelyParallel Simulator). In the process of simulating the change of crystal structure, the position change trajectory of each molecule can be tracked. By comparing the initial state and the changed state, the difference of the molecular position is calculated to determine the molecular transformation displacement. At the same time, by counting the number of molecules with obvious position changes, the number of position-transformed molecules is obtained. The critical number of position-transformed molecules and the critical molecular transformation displacement can be directly obtained from the materials science database.

[0037] In a specific embodiment, the characteristic values ​​of each crystal structure to be screened are obtained as follows: ; In the formula, represents the characteristic value of the jth crystal structure to be screened, e represents the natural constant, represents the number of position-transformed molecules in the jth crystal structure, represents the t-th molecular transformation displacement of the j-th crystal structure, Indicates the number of preset critical position transformation molecules, represents the preset critical molecular transformation displacement, Indicates the influence weight of each crystal structure characteristic value to be screened corresponding to the preset number of position transformation molecules, It represents the influence weight of the characteristic value of each crystal structure to be screened corresponding to the preset molecular transformation displacement, i represents the number of the crystal structure to be screened, j=1, 2, 3, ..., n, n represents the total number of crystal structures to be screened, t represents the number of each molecule in the crystal structure, t=1, 2, 3, ..., m, m represents the total number of crystal structures.

[0038] When performing the screening of crystal structure features, and The influence weights of the characteristic values ​​of the crystal structures to be screened corresponding to the number of position-transformed molecules and the molecular transformation displacement can be directly obtained from the materials science database. These weight values ​​respectively reflect the degree of their influence on the characteristic values ​​of the crystal structures to be screened, and there are preset mapping rules between their corresponding relationships. For example, the number of molecules in the crystal and the influence weights of the characteristic values ​​of the crystal structures to be screened corresponding to the number of position-transformed molecules and the molecular transformation displacement obtained from the materials science database form a mapping set. The number of molecules in the crystal is input into the mapping set to obtain the influence weights of the characteristic values ​​of the crystal structures to be screened corresponding to the number of position-transformed molecules and the molecular transformation displacement. The mapping method can be either one-to-one or many-to-one. In this example, the value range of the weight is limited to between 0 and 1 (excluding 0 and 1).

[0039] In this embodiment, the characteristic value of the crystal structure to be screened is used to quantitatively evaluate the stability and potential performance of the crystal structure. The fewer the number of position-changing molecules or the smaller the molecular transformation displacement, the smaller the characteristic value of the crystal structure to be screened, and the lower the stability and potential performance of the crystal structure.

[0040] The algorithm of this embodiment combines the number of position-changing molecules and the molecular transformation displacement, and comprehensively analyzes to obtain the characteristic value of the crystal structure to be screened. In this formula, the number of position-changing molecules and the molecular transformation displacement influence each other. When the crystal is affected by external or internal factors, if the number of position-changing molecules is greater, it means that more molecules have changed their positions, resulting in a more obvious overall change in the crystal structure, and the molecular transformation displacement is also greater. By comprehensively analyzing the number of position-changing molecules and the molecular transformation displacement, the characteristic value of the crystal structure to be screened can be accurately obtained, which quantitatively reflects the degree of change and complexity of the crystal structure.

[0041] Specifically, the characteristic value of each crystal structure to be screened is compared with the crystal structure characteristic threshold preset in the materials science database to obtain each target crystal structure. The specific analysis process is: if a characteristic value of a crystal structure to be screened is greater than or equal to the preset crystal structure characteristic threshold, the crystal structure to be screened is marked as a target crystal structure; if a characteristic value of a crystal structure to be screened is less than the preset crystal structure characteristic threshold, the crystal structure to be screened is marked as an unqualified target crystal structure; and each target crystal structure is obtained by statistics.

[0042] It should be understood that in the present embodiment, a preset crystal structure characteristic threshold is stored in the material science database to measure whether the crystal structure meets the standard. Compare the characteristic value corresponding to each crystal structure to be screened with the crystal structure characteristic threshold. If the characteristic value of a crystal structure to be screened is greater than or equal to the crystal structure characteristic threshold, it means that the crystal structure has reached or exceeded the established standard in terms of stability, rationality of molecular arrangement or other key performances, and the crystal structure to be screened is marked as a target crystal structure. On the contrary, if the characteristic value of a crystal structure to be screened is less than the crystal structure characteristic threshold, it indicates that the crystal structure has not reached the standard, and there are some problems such as unreasonable molecular spacing and insufficient structural stability, and the crystal structure to be screened is marked as an unqualified target crystal structure. After completing the comparison and marking process for all crystal structures to be screened, all the individuals marked as target crystal structures are counted out and summarized to form the final target crystal structure set. Each target crystal structure is a crystal sample that preliminarily meets the requirements and has further analysis or application value after screening. Each target crystal structure characteristic value is obtained by screening the characteristic value of each crystal structure to be screened.

[0043] The stability of each target crystal structure is verified to obtain a performance evaluation value of each target crystal structure. Secondary screening is performed based on the characteristic values ​​of each target crystal structure and the performance evaluation value of each target crystal structure to obtain each qualified crystal structure and output feedback.

[0044] Specifically, the stability of each target crystal structure is verified to obtain a performance evaluation value of each target crystal structure, and the specific analysis process is as follows: the stability verification includes mechanical stability verification, thermodynamic stability verification and kinetic stability verification; the stability data of each target crystal structure is obtained through stability verification, and the stability data of each target crystal structure includes the number of negative elements in the elastic constant matrix of each target crystal structure, the energy of the crystal system at each monitoring time point, and the number of imaginary frequencies of the phonon spectrum; based on the stability data of each target crystal structure, a comprehensive analysis is performed to obtain the performance evaluation value of each target crystal structure.

[0045] In this embodiment, mechanical stability is verified by the elastic constant matrix, and the crystal elastic constant matrix is ​​a symmetrical second-order tensor that contains the relationship between the internal stress and strain of the crystal when it is subjected to external forces. If all elements of the elastic constant matrix are positive, the crystal has elastic stability in all directions and will not break or deform. If there are negative elements, the crystal is unstable in the corresponding direction and is prone to breakage or deformation. Thermodynamic stability is verified by molecular dynamics simulation, and the stability of the crystal structure can be evaluated by simulating the behavior of the crystal at different temperatures and pressures. The stability of the crystal can be determined by observing the change of the energy of the crystal system over time. There is no large fluctuation, which may indicate that the crystal structure is stable. Dynamic stability is verified by phonon dispersion spectrum, and the phonon spectrum in the crystal describes the relationship between the energy and momentum of the phonons in the crystal. If there is no imaginary frequency in the phonon spectrum, that is, the frequencies of all phonons are real numbers, then it can be considered that the crystal structure is dynamically stable. If there is an imaginary frequency, it means that the crystal structure is dynamically unstable, and phase transition or structural deformation may occur. Phase transition is the process by which a substance changes from one structure to another under different conditions (such as temperature, pressure, etc.). In crystals, phase transition involves the reorganization of the atomic arrangement, which can lead to significant changes in the properties of the crystal. For example, a metal may undergo a phase transition from an ordered crystal structure to a disordered liquid structure when heated to a high enough temperature.

[0046] It should be understood that the mechanical stability verification in this embodiment is to examine the stability of the crystal structure when subjected to external forces. The elastic constant matrix reflects the response of the crystal to strain. If negative elements appear in the elastic constant matrix, it means that the crystal may be structurally unstable when subjected to force, and the number of negative elements is a key indicator for measuring mechanical stability. Thermodynamic stability verification focuses on the stability of the crystal structure under different temperature conditions. The energy of the crystal system recorded at each monitoring time point can reflect the energy state of the crystal in a thermal environment. The lower the energy, the more the crystal structure tends to be thermodynamically stable; if the system energy fluctuates abnormally or changes that do not conform to the laws of thermodynamics, it implies that there are thermodynamic instability factors in the crystal structure. Dynamic stability verification mainly analyzes the movement of atoms and molecules in the crystal. The number of imaginary frequencies of the phonon spectrum is the key basis. The phonon spectrum describes the atomic vibration mode of the crystal. Under normal circumstances, the vibration frequency should be a real number. Once an imaginary frequency appears, it means that the atomic vibration mode is unstable, which will cause the dynamic instability of the crystal structure. The number of negative elements in the elastic constant matrix of each target crystal structure can be calculated by first-principles calculation software, such as VASP (Vienna Ab-initioSimulation Package). When calculating the elastic constant matrix, VASP applies a small strain to the crystal and then calculates the corresponding stress change to obtain the elastic constant matrix. The number of negative elements can be directly counted from the calculation results. The energy of the crystal system at each monitoring time point can be calculated and recorded in real time by molecular dynamics simulation software, such as LAMMPS, in the dynamic process of simulating the crystal. The energy includes kinetic energy and potential energy (such as intermolecular potential energy, chemical bond potential energy, etc.), and the system energy value is output at different monitoring time points. The number of imaginary frequencies of the phonon spectrum can be calculated by first-principles calculation software, such as Phonopy (Phonon properties calculator). By combining with first-principles calculation software (such as VASP), the force constant matrix of the crystal is calculated, and then the phonon vibration frequency is solved using the matrix. If an imaginary frequency occurs, the software will record the number of imaginary frequencies.

[0047] Specifically, based on the stability data of each target crystal structure, a comprehensive analysis is performed to obtain a performance evaluation value of each target crystal structure. The specific analysis process is: obtaining the number of negative elements of the critical elastic constant matrix, the crystal system energy at the critical monitoring time point and the number of critical phonon spectrum imaginary frequencies from the materials science database; and processing the stability data of each target crystal structure to obtain a performance evaluation value of each target crystal structure. The target crystal structure performance evaluation value is used to quantitatively evaluate the stability of each target crystal structure.

[0048] In a specific embodiment, the method for obtaining the performance evaluation value of each target crystal structure is: ; In the formula, represents the performance evaluation value of the i-th target crystal structure, e represents the natural constant, represents the number of negative elements in the elastic constant matrix of the i-th crystal structure, represents the crystal system energy at the monitoring time point of the i-th crystal structure, represents the number of imaginary frequencies of the phonon spectrum of the i-th crystal structure, Indicates the number of negative elements in the preset critical elastic constant matrix, Indicates the energy of the crystal system at the preset critical monitoring time point, Indicates the number of preset critical phonon spectrum imaginary frequencies, Indicates the influence weight of the target crystal structure performance evaluation value corresponding to the number of negative elements in the preset elastic constant matrix, Indicates the impact weight of the target crystal structure performance evaluation value corresponding to the crystal system energy at the preset monitoring time point, represents the influence weight of the target crystal structure performance evaluation value corresponding to the preset number of phonon spectrum imaginary frequencies, i represents the number of the target crystal structure, i=1, 2, 3, ..., k, k represents the total number of target crystal structures.

[0049] When performing a target crystal structure performance evaluation, , and The weights of the target crystal structure performance evaluation values ​​corresponding to the number of negative elements of the elastic constant matrix, the energy of the crystal system at the monitoring time point, and the number of imaginary frequencies of the phonon spectrum can be directly obtained from the materials science database. These weight values ​​respectively reflect the degree of their influence on the target crystal structure performance evaluation values, and there are preset mapping rules between their corresponding relationships. For example, the number of crystal molecules and the weights of the target crystal structure performance evaluation values ​​corresponding to the number of negative elements of the elastic constant matrix, the energy of the crystal system at the monitoring time point, and the number of imaginary frequencies of the phonon spectrum obtained from the materials science database form a mapping set. The number of crystal molecules is input into the mapping set to obtain the weights of the target crystal structure performance evaluation values ​​corresponding to the number of negative elements of the elastic constant matrix, the energy of the crystal system at the monitoring time point, and the number of imaginary frequencies of the phonon spectrum. The mapping method can be one-to-one or many-to-one. In this example, the value range of the weight is limited to between 0 and 1 (excluding 0 and 1).

[0050] In this embodiment, the target crystal structure performance evaluation value is used to quantitatively evaluate the stability of each target crystal structure. The fewer the number of negative elements in the elastic constant matrix, or the smaller the crystal system energy at the monitoring time point, or the fewer the number of phonon spectrum imaginary frequencies, the larger the target crystal structure performance evaluation value is, and the higher the stability of the target crystal structure is.

[0051] The algorithm of this embodiment combines the number of negative elements of the elastic constant matrix, the crystal system energy at the monitoring time point, and the number of imaginary frequencies of the phonon spectrum, and comprehensively analyzes to obtain the target crystal structure performance evaluation value. In this formula, the number of negative elements of the elastic constant matrix, the crystal system energy at the monitoring time point, and the number of imaginary frequencies of the phonon spectrum affect each other. The number of negative elements of the elastic constant matrix directly reflects the mechanical stability of the crystal structure. When negative elements appear in the elastic constant matrix, it indicates that the crystal has structural instability when it is subjected to force. Mechanical instability will cause changes in the atomic vibration mode of the crystal, thereby affecting the number of imaginary frequencies of the phonon spectrum, because structural instability may cause abnormal modes of atomic vibration. The energy of the crystal system at the monitoring time point reflects the thermodynamic stability of the crystal structure. From an energy perspective, changes in the energy of the crystal system at the monitoring time point will affect the structural state of the crystal, and then affect the elastic constant matrix. For example, the higher the energy of the crystal system at the monitoring time point, the more negative elements will cause the crystal structure to expand or deform, affecting the elastic constant matrix. At the same time, thermodynamic instability may also intensify atomic vibrations, change the phonon spectrum, increase the number of imaginary frequencies, and affect dynamic stability. The number of imaginary frequencies of the phonon spectrum mainly reflects the dynamic stability of the crystal structure. The emergence of imaginary frequencies means that the atomic vibration mode is unstable. The fewer the number of imaginary frequencies, the more dynamically stable the crystal structure is. Dynamic stability and mechanical stability influence each other. The stability of the atomic vibration mode will affect the microstructure of the crystal, and then affect the elastic constant matrix. Moreover, dynamic instability leads to the redistribution of energy inside the crystal, causing the energy of the crystal system to change at the monitoring time point, affecting thermodynamic stability. By comprehensively analyzing the number of negative elements in the elastic constant matrix, the energy of the crystal system at the monitoring time point, and the number of imaginary frequencies in the phonon spectrum, the performance evaluation value of the target crystal structure can be accurately obtained, which quantitatively reflects the comprehensive stability of the target crystal structure in terms of mechanics, thermodynamics, and dynamics.

[0052] Specifically, secondary screening is performed based on the characteristic values ​​of each target crystal structure and the performance evaluation value of each target crystal structure. The specific analysis process is: based on the characteristic values ​​of each crystal structure to be screened and the performance evaluation value of each target crystal structure, comprehensive analysis is performed to obtain the quality evaluation index of each target crystal structure.

[0053] In a specific embodiment, the quality evaluation index of each target crystal structure is obtained as follows: ; In the formula, represents the i-th target crystal structure quality assessment index, e represents the natural constant, represents the eigenvalue of the i-th target crystal structure, represents the performance evaluation value of the i-th target crystal structure, Indicates the impact weight of the target crystal structure quality assessment index corresponding to the preset target crystal structure characteristic value, represents the target crystal structure quality evaluation index influence weight corresponding to the preset target crystal structure performance evaluation value, i represents the number of the target crystal structure, i=1, 2, 3, ..., k, k represents the total number of target crystal structures.

[0054] In a specific embodiment, Figure 2 As shown, = = 0.5, when =0.8, the functional relationship between the target crystal structure characteristic value and the target crystal structure quality evaluation index is shown in curve a; when =1.5, the target crystal structure characteristic value and the target crystal structure quality evaluation index are shown in curve b; when =2.7, the target crystal structure characteristic value and the target crystal structure quality evaluation index are shown in curve c.

[0055] When performing a target crystal structure quality assessment, and Ability to directly obtain target crystal structure characteristic values ​​and target crystal structure performance evaluation values ​​from materials science database The corresponding target crystal structure quality assessment index influence weight. These weight values ​​respectively reflect the degree of their influence on the target crystal structure quality assessment index, and there are preset mapping rules between their corresponding relationships. For example, the number of molecules in the crystal and the target crystal structure quality assessment index influence weight corresponding to the target crystal structure characteristic value and the target crystal structure performance evaluation value obtained in the materials science database form a mapping set. The number of molecules in the crystal is input into the mapping set to obtain the target crystal structure quality assessment index influence weight corresponding to the target crystal structure characteristic value and the target crystal structure performance evaluation value. The mapping method can be either one-to-one or many-to-one. In this example, the value range of the weight is limited to between 0 and 1 (excluding 0 and 1).

[0056] In this embodiment, the target crystal structure quality evaluation index is used to quantitatively evaluate the overall quality and applicability of the target crystal structure. The larger the target crystal structure characteristic value, or the larger the target crystal structure performance evaluation value, the larger the target crystal structure quality evaluation index, and the higher the overall quality and applicability of the target crystal structure.

[0057] The algorithm of this embodiment combines the target crystal structure characteristic value and the target crystal structure performance evaluation value, and comprehensively analyzes to obtain the target crystal structure quality evaluation index. In this formula, the target crystal structure characteristic value and the target crystal structure performance evaluation value influence each other. The larger the molecular displacement, the larger the target crystal structure characteristic value, which may cause more negative elements to appear in the elastic constant matrix, and the smaller the target crystal structure performance evaluation value. By comprehensively analyzing the target crystal structure characteristic value and the target crystal structure performance evaluation value, the target crystal structure quality evaluation index can be accurately obtained, which quantitatively reflects the balance and synergy between the structural optimization potential and performance stability of the target crystal structure.

[0058] If a target crystal structure quality assessment index is greater than or equal to a preset target crystal structure quality assessment index threshold, the target crystal structure is marked as a qualified crystal structure; if a target crystal structure quality assessment index is less than a preset target crystal structure quality assessment index threshold, the target crystal structure is marked as an unqualified crystal structure; statistics of each qualified crystal structure are obtained and feedback output is performed.

[0059] It should be understood that in this embodiment, the quality assessment index of each target crystal structure is compared with the target crystal structure quality assessment index threshold value preset in the materials science database. If the quality assessment index of a target crystal structure is greater than or equal to the target crystal structure quality assessment index threshold value, it means that the crystal structure meets or exceeds the established requirements in key dimensions such as structural feature rationality and performance stability, and the target crystal structure is marked as a qualified crystal structure. On the contrary, if the quality assessment index of a target crystal structure is less than the target crystal structure quality assessment index threshold value, it means that the crystal structure does not meet the requirements in terms of comprehensive performance, and the target crystal structure is marked as an unqualified crystal structure. After completing the comparison and marking operations of all target crystal structures, the crystals marked as qualified crystal structures are uniformly summarized and finally fed back and output.

[0060] Specifically, each qualified crystal structure is statistically obtained and feedback output is performed, and the specific process is: according to the quality evaluation index of each qualified crystal structure, each qualified crystal structure is sorted in order from large to small to obtain the output order of the qualified crystal structure; and each qualified crystal structure is displayed and output according to the output order of the qualified crystal structure.

[0061] It should be understood that the quality assessment index of the qualified crystal structure in the present embodiment comprehensively reflects the overall quality and applicability of the crystal structure. The higher the value, the better the performance of the crystal in terms of structural feature rationality, performance stability, etc. Therefore, the quality assessment index size of the qualified crystal structure is sorted to distinguish the degree of superiority and inferiority between different qualified crystals. All qualified crystal structures are arranged in order from large to small according to the quality assessment index, and the output order of the qualified crystal structure is determined. Such a sorting method makes the crystals with the best quality come first, which is convenient for subsequent rapid positioning and priority attention to samples with outstanding quality. According to the output order of the qualified crystal structure arranged, these qualified crystals are displayed and output one by one. The display output can be in the form of a data list, a chart or a special crystal structure visualization interface, which is convenient for users to intuitively view the screening results, quickly obtain the best quality crystal structure information, and provide a clear and orderly reference for further experimental research, material application selection and other work.

[0062] The materials science database is used to store various key data information in the field of materials science, including the number of critical position transformation molecules, critical molecular transformation displacement, the number of negative elements in the critical elastic constant matrix, the energy of the crystal system at the critical monitoring time point, and the number of critical phonon spectrum imaginary frequencies. The data in the materials science database can be obtained from professional materials experimental institutions, academic research team results data, university materials science related research project data, and large materials companies R&D test data. These data provide an important data foundation and reference basis for materials science research, crystal structure screening and analysis, and new material development.

[0063] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that technicians in the relevant technical field can understand and use the present invention well. As long as they do not deviate from the structure of the present invention or exceed the scope defined by the present invention, they should all belong to the protection scope of the present invention.

Claims

1. A crystal structure optimization application method based on a graph convolutional neural network generation model, characterized in that: include: Acquiring initial crystal structure data, and performing data augmentation on the initial crystal structure data to obtain characteristic data of each crystal structure to be screened; Training a graph convolutional neural network, screening the characteristic data of each crystal structure to be screened according to the graph convolutional neural network, obtaining each target crystal structure, and obtaining characteristic values ​​of each target crystal structure; The stability of each target crystal structure is verified to obtain a performance evaluation value of each target crystal structure. Secondary screening is performed based on the characteristic values ​​of each target crystal structure and the performance evaluation value of each target crystal structure to obtain each qualified crystal structure and output feedback.

2. The crystal structure optimization application method based on the graph convolutional neural network generation model according to claim 1 is characterized in that: The initial crystal structure data is augmented to obtain characteristic data of each crystal structure to be screened. The specific analysis process is as follows: Molecular coordinate data are extracted from the initial crystal structure data, random number variables are obtained by random extraction, and the molecular coordinate data are transformed according to the random number variables to obtain characteristic data of each crystal structure to be screened, wherein the characteristic data of each crystal structure to be screened includes the molecular coordinates of each crystal structure to be screened.

3. The crystal structure optimization application method based on the graph convolutional neural network generation model according to claim 1 is characterized in that: The specific analysis process of training the graph convolutional neural network is as follows: Set up the graph convolutional neural network and initialize the graph structure; Set up the Generative Adversarial Network, including initializing the model weights of the generator and discriminator, initializing the loss function and optimization parameters, and setting the initial training parameters of the Generative Adversarial Network; The structural feature data of each crystal to be screened are imported into the graph convolutional neural network, and the graph convolutional neural network is trained.

4. The crystal structure optimization application method based on the graph convolutional neural network generation model according to claim 1 is characterized in that: The graph convolutional neural network is used to screen the characteristic data of each crystal structure to be screened, and the specific analysis process is as follows: Obtain the number of critical position transformation molecules and critical molecular transformation displacement from the materials science database; Extracting the corresponding molecular coordinates from the initial crystal structure data and the characteristic data of each crystal structure to be screened, further obtaining the transformation displacement of each molecule in each crystal structure to be screened, and comprehensively analyzing to obtain the characteristic values ​​of each crystal structure to be screened; The characteristic values ​​of each crystal structure to be screened are compared with the crystal structure characteristic thresholds preset in the materials science database to obtain each target crystal structure.

5. The crystal structure optimization application method based on the graph convolutional neural network generation model according to claim 4 is characterized in that: The characteristic values ​​of each crystal structure to be screened are compared with the crystal structure characteristic thresholds preset in the material science database to obtain each target crystal structure. The specific analysis process is as follows: If a characteristic value of a crystal structure to be screened is greater than or equal to a preset crystal structure characteristic threshold, the crystal structure to be screened is marked as a target crystal structure; If a characteristic value of a crystal structure to be screened is less than a preset crystal structure characteristic threshold, the crystal structure to be screened is marked as an unqualified target crystal structure; The target crystal structures were statistically obtained.

6. The crystal structure optimization application method based on the graph convolutional neural network generation model according to claim 1 is characterized in that: The stability of each target crystal structure is verified to obtain the performance evaluation value of each target crystal structure. The specific analysis process is as follows: The stability verification includes mechanical stability verification, thermodynamic stability verification and kinetic stability verification; The stability data of each target crystal structure are obtained by stability verification, wherein the stability data of each target crystal structure include the number of negative elements of the elastic constant matrix of each target crystal structure, the energy of the crystal system at each monitoring time point, and the number of imaginary frequencies of the phonon spectrum; Based on the stability data of each target crystal structure, a comprehensive analysis is performed to obtain the performance evaluation value of each target crystal structure.

7. The crystal structure optimization application method based on the graph convolutional neural network generation model according to claim 6 is characterized in that: According to the stability data of each target crystal structure, a comprehensive analysis is performed to obtain the performance evaluation value of each target crystal structure. The specific analysis process is as follows: Obtain the number of negative elements in the critical elastic constant matrix, the energy of the crystal system at the critical monitoring time point, and the number of imaginary frequencies of the critical phonon spectrum from the materials science database; By processing the stability data of each target crystal structure, a performance evaluation value of each target crystal structure is obtained, and the target crystal structure performance evaluation value is used to quantitatively evaluate the stability of each target crystal structure.

8. The crystal structure optimization application method based on the graph convolutional neural network generation model according to claim 1 is characterized in that: The secondary screening is performed according to the characteristic values ​​of each target crystal structure and the performance evaluation values ​​of each target crystal structure. The specific analysis process is as follows: According to the characteristic values ​​of each target crystal structure and the performance evaluation values ​​of each target crystal structure, a quality evaluation index of each target crystal structure is obtained by comprehensive analysis; If a target crystal structure quality assessment index is greater than or equal to a preset target crystal structure quality assessment index threshold, the target crystal structure is marked as a qualified crystal structure; If a target crystal structure quality assessment index is less than a preset target crystal structure quality assessment index threshold, the target crystal structure is marked as an unqualified crystal structure; The qualified crystal structures are statistically obtained and fed back and output.

9. The crystal structure optimization application method based on the graph convolutional neural network generation model according to claim 8 is characterized in that: The statistics of each qualified crystal structure are obtained and feedback is output. The specific process is as follows: According to the quality evaluation index of each qualified crystal structure, the qualified crystal structures are sorted in descending order to obtain the output order of the qualified crystal structures; Each qualified crystal structure is displayed and output according to the output order of the qualified crystal structures.

10. The crystal structure optimization application method based on the graph convolutional neural network generation model according to claim 8, characterized in that: The method for obtaining the quality evaluation index of each target crystal structure is as follows: ； In the formula, represents the i-th target crystal structure quality assessment index, e represents the natural constant, represents the eigenvalue of the i-th target crystal structure, represents the performance evaluation value of the i-th target crystal structure, Indicates the impact weight of the target crystal structure quality assessment index corresponding to the preset target crystal structure characteristic value, represents the target crystal structure quality evaluation index influence weight corresponding to the preset target crystal structure performance evaluation value, i represents the number of the target crystal structure, i=1, 2, 3, ..., k, k represents the total number of target crystal structures.

Images