Design And Implementation Of An Integrated Platform For High Throughput Material Computing And Machine Learning

With the wide application of functional materials in industrial manufacturing,aerospace and other high-tech fields,the traditional materials research and development mode of experimental methods can not meet the needs of the market.In recent years,with the rapid development of computer science and technology and artificial intelligence theory,the cross fusion of high-throughput computational material design methods and machine learning algorithms has made remarkable progress in the research and development of new materials,speeding up the pace of material research.However,there is a certain threshold in the practical application of high-throughput computing and machine learning methods.How to reduce the difficulty of using high-throughput computing and machine learning methods in the process of material development has become a major problem to be solved in material research.In order to solve the above problem,this paper designed and implemented a high flux material calculation and machine learning integration platform,the platform system USES plug-in technology,machine learning techniques,and directing a backend database,realize the efficient computing and data analysis of the material,by using the Python language development,the platform has good portability and versatility.Firstly,the actual requirements of the material computing platform are analyzed,and the overall architecture of the platform is designed.The platform is divided into three functional modules: model building module,high-throughput computing workflow module and machine learning module.Through the call of the function modules,platform,first of all,this material model to quickly build operation of automatic generation of complex structure of the model file,load and complete model,parameter configuration,computing tasks automatically generated and automatic data processing,materials for high flux calculation,workflow fleetly the data set to the calculation of material machine learning model training,To achieve rapid prediction of material properties.Various functional modules of the platform can run stably through several practical application tests.This platform can more conveniently predict material properties by high-throughput computing and machine learning,reduce the threshold of high-throughput computing and machine learning,and improve the efficiency of new material design and development to a certain extent.