有限元模拟和神经网络相结合的喷丸处理SAE9254钢疲劳寿命预测  

Fatigue Life Prediction for SAE9254 Steel Treated by Shot Peening Based on Finite Element Simulation Combining with Neural Network

作　　者：申建国 汪舟[1,2,3] 卢伟 罗素晖 王晓丽 罗雄[4] 郑文文 汪帆星 张旭 SHEN Jianguo;WANG Zhou;LU Wei;LUO Suhui;WANG Xiaoli;LUO Xiong;ZHENG Wenwen;WANG Fanxing;ZHANG Xu(School of Automotive Engineering,Wuhan University of Technology,Wuhan 430070,China;Hubei Key Laboratory of Advanced Technology for Automotive Components,Wuhan University of Technology,Wuhan 430070,China;Hubei Collaborative Innovation Center for Automotive Components Technology,Wuhan University of Technology,Wuhan 430070,China;Guangzhou Huade Automobile Spring Co.,Ltd.,Guangzhou 511339,China)

机构地区：[1]武汉理工大学汽车工程学院,武汉430070 [2]武汉理工大学现代汽车零部件湖北省重点实验室,武汉430070 [3]武汉理工大学汽车零部件技术湖北省协同创新中心,武汉430070 [4]广州华德汽车弹簧有限公司,广州511339

出　　处：《机械工程材料》2024年第7期77-84,共8页Materials For Mechanical Engineering

基　　金：国家自然科学基金资助项目(51405356);广州华德汽车弹簧有限公司横向项目(20211h0039)。

摘　　要：采用ABAQUS有限元软件建立基于Python脚本的随机多弹丸喷丸模型,对不同弹丸直径、不同弹丸速度和不同喷丸覆盖率下喷丸处理后悬架弹簧用SAE9254钢的残余应力分布和表面粗糙度进行预测,并与试验结果进行对比;基于有限元模拟结果结合神经网络模型对试验钢的疲劳寿命进行预测,并进行试验验证。结果表明:模拟得到SAE9254钢的残余应力沿深度方向的变化曲线与试验结果吻合较好,最大残余压应力的相对误差约为14.77%,表面粗糙度的相对误差约为3.18%,建立的随机多弹丸喷丸模型能够准确地预测SAE9254钢喷丸后的残余应力分布及表面粗糙度。采用有限元模拟与神经网络相结合的方法得到的疲劳寿命预测值和试验值的平均相对误差为6.85%,该方法可以准确地预测SAE9254钢的疲劳寿命。A stochastic multiple shot peening model based on Python scripts was established by ABAQUS finite element software.The residual stress distribution and surface roughness of SAE9254 steel for suspension springs after shot peening under different shot diameters,different shot velocities,and different shot peening coverage rates were predicted and compared with test results.Based on the finite element simulation results and neural network model,the fatigue life of the test steel was predicted,and experimental verification was carried out.The results show that the simulated curves of residual stress along the depth direction of SAE9254 steel were in good agreement with the test results,the relative error of the maximum residual compressive stress was about 14.77%,and the relative error of surface roughness was about 3.18%,which indicated that the established stochastic multiple shot peening model could accurately predict the residual stress distribution and surface roughness of SAE9254 steel after shot peening.The average relative error between the fatigue life prediction values obtained by the method combining finite element simulation and neural network and the experimental values was 6.85%,indicating that this method could accurately predict the fatigue life of SAE9254 steel.

关 键 词：SAE9254钢 喷丸 表面粗糙度 有限元模拟 神经网络 疲劳寿命 

分 类 号：TG178[金属学及工艺—金属表面处理]