Cam-followers provide reliable and controlled motions in various mechanical systems. Due to the highly fluctuating load between the cam and follower in operation, the cam-follower may be subjected to a high risk of co...Cam-followers provide reliable and controlled motions in various mechanical systems. Due to the highly fluctuating load between the cam and follower in operation, the cam-follower may be subjected to a high risk of contact fatigue failure. This paper assesses the fatigue life of a cycloidal displacement cam and a flat-faced follower under the defined loads and constraints. Computer-aided design (CAD) model of the cam-follower is developed in CATIA software and imported to ANSYS software for finite element analysis (FEA) of fatigue life. MATLAB programming is developed for determining the appropriate spring constant and pre-load force to always keep the cam and follower in contact. The fatigue life of the cam-follower has been estimated under the specified operating conditions. The analysis method can be applied to investigate the fatigue life of cams with other profiles, including the modified trapezoidal functions, polynomial functions, etc.展开更多
针对某直列四缸增压柴油机排气歧管在台架耐久试验中出现开裂现象,应用有限元(finite element analysis,FEA)-计算流体力学(computational fluid dynamics,CFD)耦合方法对排气歧管进行了热应力分析。排气歧管热应力分析模型中的材料性...针对某直列四缸增压柴油机排气歧管在台架耐久试验中出现开裂现象,应用有限元(finite element analysis,FEA)-计算流体力学(computational fluid dynamics,CFD)耦合方法对排气歧管进行了热应力分析。排气歧管热应力分析模型中的材料性质随着温度变化的关系由试验测定;在相应发动机工况下对模型预测的温度场和台架上测量的排气歧管温度进行对比,以对模型进行标定。应用标定后的模型分析排气歧管在给定热负荷条件下的应力-应变分布。FEACFD耦合分析结果表明:发生开裂的区域为高塑性应变区,从而推论导致排气歧管开裂原因为热力耦合应力作用下产生的塑性变形,即失效形式为热机械疲劳。以等效塑性应变作为塑性变形的度量及许可的等效塑性应变经验值为判据对排气歧管进行了设计改进,并从三种改进方案中找出最安全的方案进行试验验证,改进后的排气歧管顺利通过发动机台架耐久试验考核。展开更多
Based on finite element analysis of thermal mechanical behavior, structural optimization design was proposed for a side cooling collimating mirror subjected to high heat load for a beamline at SSRF(Shanghai Synchrotro...Based on finite element analysis of thermal mechanical behavior, structural optimization design was proposed for a side cooling collimating mirror subjected to high heat load for a beamline at SSRF(Shanghai Synchrotron Radiation Facility). The temperature distribution,stress concentration effect, maximum equivalent(vonMises) stress, and slope error of the mirror were analyzed.In particular, the cooling water channels of the traditional structural design were optimized, and the modified designs were further optimized. Although the traditional structural and the improved designs could meet requirements for the temperature and thermal stress, the deformation gradients were relatively large for several structural designs, and this led to larger slope error. The further improved structural designs could be of better performance.展开更多
目的针对图像分类任务中对于细粒度特征提取困难,同时背景噪声和不相关区域影响网络对目标特征学习的问题,本文提出随机空洞卷积的图像分类网络(image classification network with random dilated convolution,RDCNet)。方法RDCNet网络...目的针对图像分类任务中对于细粒度特征提取困难,同时背景噪声和不相关区域影响网络对目标特征学习的问题,本文提出随机空洞卷积的图像分类网络(image classification network with random dilated convolution,RDCNet)。方法RDCNet网络以ResNet-34(residual network-34)为基线网络。首先,提出多分支随机空洞卷积(multi-branch random dilated convolution,MRDC)模块,通过多个分支的卷积操作和随机膨胀卷积核的设计,实现了从不同尺度和感受野上对细粒度特征的有效捕捉。通过引入细粒度特征增强(fine-grained feature enhancement,FGFE)模块,实现对全局信息的学习和局部特征的增强,提升了网络局部特征提取和全局上下文理解能力。同时引入随机掩码机制动态地遮蔽部分输入特征和卷积核权重,不仅可以通过多样化的特征组合学习更加健壮和鲁棒性的表示,还能够有效减少过拟合,提升对噪声和不相关区域的适应能力。最后,提出上下文激励(context excitation,CE)模块,通过引入上下文信息并动态调整特征通道的权重,增强网络对关键特征的关注能力,抑制背景噪声的干扰,提升了特征的表达能力。结果本文方法在CIFAR-10(Canadian institute for advanced research 10)、CIFAR-100、SVHN(street view house number)、Imagenette和Imagewoof数据集上均有良好的分类准确率,相比于性能第2的模型,分类准确率分别提高了0.02%、1.12%、0.18%、4.73%和3.56%。实验结果表明,RDCNet具有较高的分类性能。结论随机空洞卷积的图像分类网络具有更强的细粒度特征敏感度,能够在多尺度和上下文中提取丰富的特征信息,较好地关注关键特征,对复杂背景下目标具有更优秀的辨识能力,从而在分类任务中表现出优秀的分类性能。展开更多
文摘Cam-followers provide reliable and controlled motions in various mechanical systems. Due to the highly fluctuating load between the cam and follower in operation, the cam-follower may be subjected to a high risk of contact fatigue failure. This paper assesses the fatigue life of a cycloidal displacement cam and a flat-faced follower under the defined loads and constraints. Computer-aided design (CAD) model of the cam-follower is developed in CATIA software and imported to ANSYS software for finite element analysis (FEA) of fatigue life. MATLAB programming is developed for determining the appropriate spring constant and pre-load force to always keep the cam and follower in contact. The fatigue life of the cam-follower has been estimated under the specified operating conditions. The analysis method can be applied to investigate the fatigue life of cams with other profiles, including the modified trapezoidal functions, polynomial functions, etc.
文摘针对某直列四缸增压柴油机排气歧管在台架耐久试验中出现开裂现象,应用有限元(finite element analysis,FEA)-计算流体力学(computational fluid dynamics,CFD)耦合方法对排气歧管进行了热应力分析。排气歧管热应力分析模型中的材料性质随着温度变化的关系由试验测定;在相应发动机工况下对模型预测的温度场和台架上测量的排气歧管温度进行对比,以对模型进行标定。应用标定后的模型分析排气歧管在给定热负荷条件下的应力-应变分布。FEACFD耦合分析结果表明:发生开裂的区域为高塑性应变区,从而推论导致排气歧管开裂原因为热力耦合应力作用下产生的塑性变形,即失效形式为热机械疲劳。以等效塑性应变作为塑性变形的度量及许可的等效塑性应变经验值为判据对排气歧管进行了设计改进,并从三种改进方案中找出最安全的方案进行试验验证,改进后的排气歧管顺利通过发动机台架耐久试验考核。
基金supported by the National Natural Science Foundation of China(No.11175243)
文摘Based on finite element analysis of thermal mechanical behavior, structural optimization design was proposed for a side cooling collimating mirror subjected to high heat load for a beamline at SSRF(Shanghai Synchrotron Radiation Facility). The temperature distribution,stress concentration effect, maximum equivalent(vonMises) stress, and slope error of the mirror were analyzed.In particular, the cooling water channels of the traditional structural design were optimized, and the modified designs were further optimized. Although the traditional structural and the improved designs could meet requirements for the temperature and thermal stress, the deformation gradients were relatively large for several structural designs, and this led to larger slope error. The further improved structural designs could be of better performance.
文摘为确定变参数桥梁最优内力,针对第十四届全国大学生结构设计竞赛赛题中的模型进行理论分析与优化,建立单目标线性优化设计数学模型和桥梁结构简化计算模型.采用穷举算法,结合Visual C++编程优化计算,其中包括桥梁主跨跨径的优化、加载点荷载值选择,进行静力分析、结构优化设计和实际模型试验.推导了数值计算公式,提出以弯曲应变能最小为目标的桥梁跨径、荷载加载位置等参数随机优化的方法,寻求在荷载作用下结构的竖向位移和内力的最小值,得到荷载布置方式,反算主跨跨径,利用有限元软件建模分析,并进行试验验证,得到布载方式1为最优布载,P 1~P 8值分别为40、50、120、130、60、70、80、90 N.
文摘目的针对图像分类任务中对于细粒度特征提取困难,同时背景噪声和不相关区域影响网络对目标特征学习的问题,本文提出随机空洞卷积的图像分类网络(image classification network with random dilated convolution,RDCNet)。方法RDCNet网络以ResNet-34(residual network-34)为基线网络。首先,提出多分支随机空洞卷积(multi-branch random dilated convolution,MRDC)模块,通过多个分支的卷积操作和随机膨胀卷积核的设计,实现了从不同尺度和感受野上对细粒度特征的有效捕捉。通过引入细粒度特征增强(fine-grained feature enhancement,FGFE)模块,实现对全局信息的学习和局部特征的增强,提升了网络局部特征提取和全局上下文理解能力。同时引入随机掩码机制动态地遮蔽部分输入特征和卷积核权重,不仅可以通过多样化的特征组合学习更加健壮和鲁棒性的表示,还能够有效减少过拟合,提升对噪声和不相关区域的适应能力。最后,提出上下文激励(context excitation,CE)模块,通过引入上下文信息并动态调整特征通道的权重,增强网络对关键特征的关注能力,抑制背景噪声的干扰,提升了特征的表达能力。结果本文方法在CIFAR-10(Canadian institute for advanced research 10)、CIFAR-100、SVHN(street view house number)、Imagenette和Imagewoof数据集上均有良好的分类准确率,相比于性能第2的模型,分类准确率分别提高了0.02%、1.12%、0.18%、4.73%和3.56%。实验结果表明,RDCNet具有较高的分类性能。结论随机空洞卷积的图像分类网络具有更强的细粒度特征敏感度,能够在多尺度和上下文中提取丰富的特征信息,较好地关注关键特征,对复杂背景下目标具有更优秀的辨识能力,从而在分类任务中表现出优秀的分类性能。
