基于神经元网络的薄壁筒滚珠旋压成形缺陷诊断  

Diagnosis of defects in ball spinning deformation of thin-walled tubular part based on ANN

作　　者：江树勇[1] 薛克敏[2] 李春峰[3] 张军[1] 

机构地区：[1]哈尔滨工程大学工程训练中心,黑龙江哈尔滨150001 [2]合肥工业大学材料科学与工程学院,安徽合肥230009 [3]哈尔滨工业大学材料科学与工程学院黑龙江哈尔滨,150001

出　　处：《锻压技术》2006年第3期79-83,共5页Forging & Stamping Technology

摘　　要：作为一种连续局部塑性成形工艺,滚珠旋压被应用于制造高强度、高精度的纵向内筋薄壁筒形件。通过使用铝合金作为旋压材料,在实验的基础上分析了滚珠旋压过程中金属材料非稳定流动的基本原理及旋压件产生表面质量缺陷的原因。以人工神经元网络为基础,对旋压件的表面质量缺陷进行了预测。实验证明,神经元网络能够精确地诊断旋压件的表面质量缺陷。As a successively and locally plastic deformation process, ball spinning is applied in order to manufacture high-strength and high-precision thin-walled tubular part with longitudinal inner ribs. By using aluminum alloy as spinning material, based on the experiments, not only the basic principle with respect to non-steady flow of metal material in ball spinning, but also the reasons for surface quality defects of the spun parts are analyzed. On the basis of artificial neural networks （ANN）, the surface quality defects of the spun parts are predicted. Experiments have proved that ANN can predict and diagnose the surface quality defects of the spun part successfully.

关 键 词：滚珠旋压 强力旋压 神经元网络 铝合金 

分 类 号：TG376[金属学及工艺—金属压力加工]