5 results about "Crystal properties" patented technology

The application relates to a training method and device of a crystal property prediction model and a computer device. The method comprises the following steps: obtaining a sample crystal and a property label corresponding to the sample crystal; the sample crystal comprises a plurality of sample atoms; determining respective atomic coordinates of each sample atom, and determining orbital characteristics of the sample atoms according to the atomic coordinates; determining respective atomic potential energies of the sample atoms according to the orbital characteristics, and determining a predicted property corresponding to the sample crystal according to the atomic potential energies; training the crystal property prediction model according to the difference between the predicted property and the property label, stopping when a first training stop condition is reached, and obtaining a trained crystal property prediction model; and the trained crystal property prediction model is used to determine the property of the crystal. The method can improve the accuracy of predicting the property of the crystal.

The application discloses a device for micro in-situ characterization of hydrate crystal characteristics of a solid-liquid interface, which comprises a visual reaction kettle, a test sample, a simulated disturbance system, a confocal Raman spectrometer and a constant-temperature control system. The test sample comprises solid wafers arranged oppositely and liquid bridges between the solid wafers. The test sample is placed into the visual reaction kettle, and the temperature and pressure in the visual reaction kettle are adjusted to a low-temperature and high-pressure state. According to different test requirements, a three-axis moving force arm of the simulated disturbance system can drive the solid wafers to move in different directions, so that different degrees of disturbance can be simulated. Meanwhile, the disturbance of fluid injection / extraction, a magnetic field, an electric field and pH value change can also be simulated. In the state of disturbance, micro-analysis information of hydrate structure, components, pore occupancy, hydration number and the like in a natural gas hydrate generation / dissolution dynamic process on the solid-liquid interface can be obtained, so as to provide a theoretical basis for natural gas hydrate accumulation and exploitation.

The present application relates to the technical field of crystal material property prediction, and in particular to a crystal property prediction method and model based on graph convolution and attention mechanism, which first one-hot encodes and projects atomic key physical properties as node features, then calculates local and global potential edge features and fuses them, combines atomic geometric information to generate directional edge features through a directional message passing network, inputs the multi-source edge features and node features into a graph convolution network to complete feature iterative updating, dynamically filters the features through element-by-element gating attention mechanism, obtains a crystal-level global representation through global pooling, and inputs the crystal-level global representation into a fully connected network to output a prediction result. The model corresponds to three major modules of input, interaction and output, can accurately model atomic multi-scale interactions, explicitly represent directional interactions, dynamically adjust feature weights, has excellent prediction accuracy and generalization ability, and is suitable for property prediction of various crystal systems.

The invention relates to the technical field of artificial intelligence, and discloses a crystal structure screening method and system, a terminal and a storage medium. The method comprises the following steps: generating a plurality of first candidate crystal structures through a crystal structure generation model according to input target crystal property parameters; performing physicochemical property prediction on the plurality of first candidate crystal structures to obtain a plurality of candidate crystal property parameters; screening out a second candidate crystal structure meeting a preset condition from the plurality of first candidate crystal structures according to the difference between the plurality of candidate crystal property parameters and the target crystal property parameter; and taking the second candidate crystal structure and the target crystal property parameters as training data, updating model parameters of the crystal structure generation model, repeatedly executing the generation, prediction and screening steps, and carrying out iterative optimization until a preset termination condition is met, thereby obtaining a plurality of target crystal structures. According to the method, the candidate structure related to the target property can be automatically generated, and the design efficiency is improved.

This invention relates to the field of crystal property prediction technology, and proposes a method for predicting the properties of crystal materials. The method includes: acquiring a crystal informatics file of a target crystal; mapping the atoms in the target crystal to a three-dimensional grid with adjustable resolution based on the atomic structure information in the crystal informatics file, and filling each grid cell with the feature values ​​of the atoms at corresponding positions to obtain the three-dimensional distribution features of each atom; inputting the three-dimensional distribution features of each atom into a neural network to obtain the property prediction result of the target crystal, wherein the neural network is trained with the three-dimensional distribution features of all atoms in the known crystal as input and property labels as supervision. This invention achieves end-to-end direct prediction of crystal properties from a crystal informatics file by reconstructing the crystal structure into three-dimensional distribution features, possessing the advantages of high computational efficiency, complete preservation of spatial information, and flexible and adjustable representation granularity.