The concave die design of self-pierce riveting(SPR) is of critical importance for product quality. The optimization of concave die parameters based on orthogonal test is proposed to explore the relationship between se...The concave die design of self-pierce riveting(SPR) is of critical importance for product quality. The optimization of concave die parameters based on orthogonal test is proposed to explore the relationship between self-pierce riveted joint quality and die parameters. There are nine independent die parameter factors in orthogonal test and each factor has 4 levels. In order to evaluate the interlock and neck thickness, we carry out numerical simulations by the software DEFORM-2D. Then, the primary and secondary factors that affect the joint quality have been found out by means of range analysis. Finally, an optimization scheme is brought forward to design concave die in SPR process, which indicates that the joint has higher quality than that of former orthogonal tests.This work can be extended by a detailed mechanical and fatigue analysis for the joint quality of SPR process.展开更多
自冲铆接技术适合铝钢等异种材料连接,接头性能可靠,在航空工业中有着广阔的应用场景,但目前针对自冲铆接缺陷无损检测的相关研究较少。提出了基于深度学习的自冲铆接偏铆缺陷检测算法,首先通过剪切力学性能试验得出偏铆自冲铆接件相较...自冲铆接技术适合铝钢等异种材料连接,接头性能可靠,在航空工业中有着广阔的应用场景,但目前针对自冲铆接缺陷无损检测的相关研究较少。提出了基于深度学习的自冲铆接偏铆缺陷检测算法,首先通过剪切力学性能试验得出偏铆自冲铆接件相较于正常铆接件力学性能下降了5.6%;然后通过自冲铆接偏铆件外部形貌特征将偏铆程度由外部特征定义在0~10的区间;最后探究了单步检测同双步检测间的检测效果差异,提出了YOLOv5s(You Only Look Once v5s)加ResNet18的检测方案,并通过Grad-CAM(Gradient-Weighted Class Activation Mapping)对不同检测方案的效果差异进行了可视化的解释。测试表明,提出的YOLOv5s加ResNet18的检测方案在所采集的数据测试集中可以达到100%正确率,高于仅用YOLOv5s取得的95.18%正确率,远高于仅用ResNet18取得的84.1%正确率。展开更多
基金the National Natural Science Foundation of China(No.51375282)the China Postdoctoral Science Foundation(No.2012T50621)the Open Fund of Shanghai Key Laboratory of Digital Manufacture for Thin-walled Structures(No.2011003)
文摘The concave die design of self-pierce riveting(SPR) is of critical importance for product quality. The optimization of concave die parameters based on orthogonal test is proposed to explore the relationship between self-pierce riveted joint quality and die parameters. There are nine independent die parameter factors in orthogonal test and each factor has 4 levels. In order to evaluate the interlock and neck thickness, we carry out numerical simulations by the software DEFORM-2D. Then, the primary and secondary factors that affect the joint quality have been found out by means of range analysis. Finally, an optimization scheme is brought forward to design concave die in SPR process, which indicates that the joint has higher quality than that of former orthogonal tests.This work can be extended by a detailed mechanical and fatigue analysis for the joint quality of SPR process.
文摘自冲铆接技术适合铝钢等异种材料连接,接头性能可靠,在航空工业中有着广阔的应用场景,但目前针对自冲铆接缺陷无损检测的相关研究较少。提出了基于深度学习的自冲铆接偏铆缺陷检测算法,首先通过剪切力学性能试验得出偏铆自冲铆接件相较于正常铆接件力学性能下降了5.6%;然后通过自冲铆接偏铆件外部形貌特征将偏铆程度由外部特征定义在0~10的区间;最后探究了单步检测同双步检测间的检测效果差异,提出了YOLOv5s(You Only Look Once v5s)加ResNet18的检测方案,并通过Grad-CAM(Gradient-Weighted Class Activation Mapping)对不同检测方案的效果差异进行了可视化的解释。测试表明,提出的YOLOv5s加ResNet18的检测方案在所采集的数据测试集中可以达到100%正确率,高于仅用YOLOv5s取得的95.18%正确率,远高于仅用ResNet18取得的84.1%正确率。
