期刊文献+

基于灰色理论和神经网络的自冲铆接头力学性能预测 被引量:16

Prediction of mechanical property of self-piercing riveted joints based on grey theory and neural network
下载PDF
导出
摘要 为研究铆接参数对自冲铆接头力学性能的影响,对AA5052板进行自冲铆接正交试验,采用灰色关联度分析了各铆接参数对接头强度的影响程度,并通过BP神经网络建立了铆接参数与接头强度的神经网络预测模型。利用获得的正交试验数据对网络进行训练,并采用未参加训练的样本验证该预测模型的有效性。结果表明:刺穿压强和整形压强对接头强度影响最大,其中刺穿压强对接头强度起决定作用,适当增大刺穿压强可以提高接头强度。采用LM算法优化的BP神经网络模型具有良好的泛化能力和预测精度,对未参加训练样本预测精度高,其预测最大相对误差为5.48%,预测精度超过90%。 To study the influence of the riveting parameters on the mechanical properties of self-piercing riveted joints,the orthogonal test of self-piercing riveting(SPR) was performed for AA5052 sheets. The influence degree of each parameter on the strength of the joints was analyzed by grey relational grade theory. Based on BP neural network,the strength prediction model of the joints related to the riveting parameters was established. The network model was trained by the experimental data,and the validity of the prediction model was verified by the untrained samples. The results show that the riveting pressure and the compressing pressure have the greatest influence on the joint strength. The riveting pressure plays a decisive role in the joint strength,and the joint strength can be improved by increasing the riveting pressure properly. The BP neural network optimized by LM algorithm has good generalization ability and prediction accuracy,the prediction accuracy of the untrained samples is high,and the maximum relative error is 5. 48%,the prediction accuracy is over 90%.
出处 《塑性工程学报》 CAS CSCD 北大核心 2017年第4期71-76,共6页 Journal of Plasticity Engineering
基金 国家自然科学基金资助项目(51565023 51565022)
关键词 自冲铆接 BP神经网络 力学性能 灰色关联度 预测 self-piercing riveting BP neural network mechanical property grey relational grade prediction
  • 相关文献

参考文献7

二级参考文献58

共引文献123

同被引文献122

引证文献16

二级引证文献50

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部