Method and system for automatically constructing quantitative structure-pharmacokinetics relationship (QSPR) model of material

Abstract

The invention relates to a method and a system for automatically constructing a QSPR model of a material. According to the method, a machine learning method is utilized; based on a data set formed bythe crystal structures of a target material system and target property data, a QSPR model of the material is trained in an automatic loop iteration manner; the target material system is subjected to crystal structure regulation and control, so structure sample space is enriched; crystal structure properties generated after regulation and control can be automatically calculated through first principles, and a training set is expanded; and the QSPR model is further trained and testing is further conducted until a test result of the QSPR model reaches preset precision or the crystal structure sample space is exhaustive. The method is characterized in that sample data is automatically generated, self-learning is carried out, and a structure-property prediction model meeting certain predictionprecision is automatically trained.

Description

technical field [0001] The invention relates to the fields of material genetic engineering and material informatics, in particular to a method and system for automatically constructing a quantitative structural property model of materials. Background technique [0002] In the field of machine learning and materials informatics, the "quantitative structural property model" model of materials, also known as the QSPR model, can effectively assist the design of new materials, but the lack of material data and the multi-disciplinary cross-cutting involved in model construction have increased its complexity. Difficulty, there is no method and system that can automatically realize the construction of a "quantitative structural property model" of materials in the prior art. SUMMARY OF THE INVENTION [0003] The purpose of the present invention is to provide a method and system for automatically constructing a quantitative structural property model of a material, which can automati...

AI Technical Summary

Problems solved by technology

[0002] In the field of machine learning and materials informatics, the "quantitative structural property model" model of materials, also known as the QSPR model, can effectively assist the design of new materials, but the lack of material data and the multidisciplinary involvement of model construction have increased its Difficulty, there is no method and system in the prior art that can automatically realize the construction of "quantitative structural property model" of materials

Images