基于物理信息数据驱动方法的TC4合金增材制造温度场预测  

Prediction of temperature field in additive manufacturing of TC4 alloy based on physics⁃informed data⁃driven approach

作　　者：刘鹏伟 廖福渝 赵英杰 赵紫松 宋立军[2] 刘鑫刚[1] LIU Peng-wei;LIAO Fu-yu;ZHAO Ying-jie;ZHAO Zi-song;SONG Li-jun;LIU Xin-gang(School of Mechanical Engineering,Yanshan University,Qinghuangdao 066004,China;College of Mechanical and Vehicle Engineering,Hunan University,Changsha 410082,China)

机构地区：[1]燕山大学机械工程学院,河北秦皇岛066004 [2]湖南大学机械与运载工程学院,湖南长沙410082

出　　处：《塑性工程学报》2023年第6期166-175,共10页Journal of Plasticity Engineering

基　　金：国家自然科学基金资助项目(12202375);河北省引进留学人员资助项目(C20210328);燕山大学基础创新科研培育项目(2021LGQN029)。

摘　　要：通过整合有限元热传导模型和长短时记忆网络,提出了一种基于物理信息的数据驱动方法,并应用于TC4合金直接能量沉积过程的温度场预测研究。该方法以有限元计算的TC4合金直接能量沉积过程中前n个时刻的2D/3D温度场图像和部分实验数据为训练集,预测第n+1时刻的温度场,从而实现直接能量沉积过程的温度场演化模拟。为了保证预测结果的可靠性,采用验证集方法对数据驱动计算模型进行测试,适当的实验数据对模型进行校正。结果表明,提出的数据驱动计算模型能够准确地预测熔池中心的平均温度和打印件的实时温度分布,此外与传统物理模型相比,其计算效率得到极大提升。A physics⁃informed data⁃driven method was developed by integrating the finite element(FE)heat transfer model and the long short⁃term memory network(LSTM),and then it was employed for the temperature field prediction investigation of the direct energy depo⁃sition(DED)process of TC4 alloy.The 2D/3D temperature field images of the firstnmoments in DED process of TC4 alloy obtained from the FE calculation and some of the experimental data were used as the training set to predict the temperature distribution atn+1 mo⁃ment,and then the temperature field evolution simulation of DED process was realized.To ensure the reliability of the prediction results,a validation set method was used to test the data⁃driven computational model,and appropriate experimental data was used to calibrate the model.The results show that the developed data⁃driven calculation model can accurately predict the average temperature of the center for melt pool and the real⁃time temperature distribution of printed parts.In addition,the calculation efficiency is substantially improved com⁃pared with the traditional physics models.

关 键 词：数据驱动 长短时记忆网络 有限元法 增材制造 温度场预测 

分 类 号：TG44[金属学及工艺—焊接]