神经网络在模具型腔温升预测中的应用  

Application of BP neural network in prediction of temperature rise for die cavity

作　　者：王雷刚[1] 孙宪萍[1] 黄瑶[1] 

机构地区：[1]江苏大学材料科学与工程学院,镇江212013

出　　处：《塑性工程学报》2008年第2期80-83,共4页Journal of Plasticity Engineering

基　　金：国家自然科学基金资助项目(50575097);江苏大学高级人才基金资助项目(04JDG037)。

摘　　要：挤压成形过程中由于坯料和模具之间的滑动接触摩擦和坯料的塑性变形产生热而使模具型腔表面温度升高,加剧模具的磨损。采用热力耦合有限元法计算挤压成形过程中模具型腔表面的温升,将模拟结果与人工神经网络相结合,以有限元模拟结果作为学习样本,训练BP神经网络模型,以此模型预测模具型腔表面的温升。根据预测结果分析了挤压锥角、挤压速度和摩擦系数对型腔温升的影响,为进一步建立挤压成形过程中模具型腔表面的温升模型和磨损预测模型奠定了基础。During the extrusion process, the heat gencrated by sliding contact friction and plastic deformation of billet leads to the temperature rise of cavity surface, which consequently aggravates the wear of dies. The temperature rise of die cavity surface during extrusion process is calculated by thermal-mechanical coupling FEM. Combining the simulation results and artificial neural network, BP neural network is trained with the simulation results as learning samples. The model is used to predict the temperature rise of die cavity surface and the influences of extrusion cone angle, extrusion speed and friction coefficient on temperature rise can be analyzed using the predictive results, which lays the foundation for the establishment of temperature rise model and wear prediction model of die cavity surface.

关 键 词：挤压模具 温升 神经网络 热力耦合 

分 类 号：TG315[金属学及工艺—金属压力加工]