Numerical Simulation And Regression Analysis Of SLM Forming Of AlCoCrFeNi_2.1 High Entropy Alloy

With the birth of high entropy alloy,the excellent properties of many elements can be displayed in one alloy material at the same time.The properties of alloy materials are not only determined by the chemical composition and internal microstructure,but also by the manufacturing technology.Additive manufacturing technology has the advantages of no need to design the mold,rapid prototyping for complex structure and less material loss.Selective Laser Melting（SLM）is one of the most important technologies in the additive manufacturing of alloys.The selection of process parameters is an important factor affecting the quality of SLM.Therefore,this paper studies the influence of SLM molding process parameters on molding quality through numerical simulation,and establishes the relational model between process parameters and simulation results through regression analysis,and obtains a regression model with high accuracy,which greatly improves the efficiency of process parameter optimization.The main research contents and conclusions are presented as follows:（1）By selecting the AlCoCrFeNii_2.1high entropy alloy with excellent as-cast properties,the SLM forming experiment of block AlCoCrFeNii_2.1high entropy alloy is completed.The thermophysical parameters of AlCoCrFeNii_2.1for numerical simulation are obtained through the experimental method.The experiment mainly includes the measurement of specific heat capacity by DSC,the measurement of thermal diffusivity by flash method,the measurement of thermal expansion coefficient by thermomechanical analysis,and the estimation of elastic modulus at high temperature by mixing method and phenomenological relationship.（2）Based on the birth-death element method and the inherent strain theory,the numerical simulation of SLM experiment is completed by constructing a two-level model of heat source-structural member,and the accuracy of the numerical simulation method is verified.The effects of laser width,laser absorption efficiency,laser power,scanning speed and powder layer thickness on the maximum total deformation of SLM are analyzed.The results show that the influence of different process parameters on the maximum total deformation is different,which meets the requirements of regression analysis algorithm for variables.（3）Taking five kinds of process parameters as independent variables and the maximum total deformation as dependent variables,500 groups of data sets established by different process parameters and maximum deformation are obtained by using random number algorithm,and the data are normalized.Five kinds of machine learning regression algorithms such as linear regression model,decision tree regression model,random forest regression model,k-nearest neighbor（KNN）regression model and support vector regression（SVR）model are used to analyze the regression effect of the five regression models.The results of laser selective melting show that the goodness of fit of SVR model with RBF as kernel function can reach 0.977,which is closest to 1,and the mean square error is the smallest,and the learning efficiency and regression effect are relatively optimal.（4）Firefly algorithm and particle swarm optimization are used to optimize the penalty factor and kernel function parameters of support vector regression model.The results show that the two algorithms have positive effects on improving the goodness of fit of SVR model,and the particle swarm optimization algorithm can make the goodness of fit of SVR model reach 0.984,which is better than the firefly algorithm.Therefore,the SVR regression model based on particle swarm optimization algorithm has good accuracy and adaptability for predicting the maximum deformation of AlCoCrFeNii_2.1block in SLM process.