Performance Prediction And Composition Design Of Rare Earth Magnesium Alloys With High Strength And High Thermal Conductivity Based On Machine Learning

With the rapid development of new energy vehicles,electronic products and aerospace industries,magnesium alloy materials with light weight,excellent mechanical properties and high thermal conductivity have attracted wide attention.Increasing the use of magnesium alloys in new energy vehicles and aerospace equipment can reduce product weight and save energy.In this paper,a compositionprocessing-property machine learning prediction model was established to accurately predict the tensile strength,yield strength,elongation and thermal conductivity of rareearth magnesium alloys.The composition and processing technology of a novel high strength and high thermal conductivity rare earth magnesium alloy were reversely designed based on machine learning model.Based on the machine learning model,the prediction model of magnesium alloy components-processing-mechanical properties was established.The components of magnesium alloy include Zn,Gd,Y,Nd,Zr and Ce,and the processing processes in magnesium alloy include extrusion,solution and aging.By comparing the root mean square error,mean absolute percentage error and deterministic coefficient of support vector machine model,BP neural network model and generalized regression neural network model,it is determined that support vector machine model has the best prediction effect.Based on the support vector machine model,the tensile strength,yield strength and elongation of magnesium alloy were accurately predicted.The decisive coefficient of the predicted tensile strength model was 0.941,that of the predicted yield strength model was 0.94,and that of the predicted elongation model was 0.79.Based on the support vector machine model,the law of composition changing the tensile strength,yield strength and elongation of magnesium alloy was investigated.Based on the idea of classification before prediction,a machine learning model which can accurately predict the thermal conductivity of rare earth magnesium alloys is established.The input values of the machine learning model include magnesium alloy components: Al,La,Zn,Y,Zr,Mn,Ce,Nd and Gd.Magnesium alloy processing technology: extrusion temperature,extrusion ratio,solution temperature,solution time,aging temperature,aging time and magnesium alloy thermal conductivity measurement temperature.First,magnesium alloy data were divided into three categories based on cluster classification.Then,root-mean-square error,mean absolute percentage error and deterministic coefficient of different categories were calculated based on 10-times cross validation,and the most appropriate machine learning model was selected for each category.Compared with a single machine learning model,the deterministic coefficient of BP neural network model is 0.88,that of generalized regression neural network model is 0.92 and that of support vector machine model is 0.92.The deterministic coefficient of the machine learning model based on classification before prediction is 0.93,which is better than the other three models.Based on the combination of machine learning model and genetic algorithm optimization,magnesium alloys with high tensile strength,high yield strength,high elongation and high thermal conductivity are recommended.In one of the recommended a new type of rare earth magnesium alloy with the highest tensile strength is 575 MPa,the highest yield strength is 532.9 MPa,the elongation of the highest is 27%,the highest thermal conductivity is 136 W/(m·K).Based on the machine learning model,the components and processing processes of rare earth magnesium alloys with both high yield strength and high elongation are recommended,and the components and processing processes of rare earth magnesium alloys with high thermal conductivity and high yield strength are recommended.