This study proposes a general imperfect thermal contact model to predict the thermal contact resistance at the interface among multi-layered composite structures.Based on the Green-Lindsay(GL)thermoelastic theory,semi...This study proposes a general imperfect thermal contact model to predict the thermal contact resistance at the interface among multi-layered composite structures.Based on the Green-Lindsay(GL)thermoelastic theory,semi analytical solutions of temperature increment and displacement of multi-layered composite structures are obtained by using the Laplace transform method,upon which the effects of thermal resistance coefficient,partition coefficient,thermal conductivity ratio and heat capacity ratio on the responses are studied.The results show that the generalized imperfect thermal contact model can realistically describe the imperfect thermal contact problem.Accordingly,it may degenerate into other thermal contact models by adjusting the thermal resistance coefficient and partition coefficient.展开更多
Damage detection in structures is performed via vibra-tion based structural identification. Modal information, such as fre-quencies and mode shapes, are widely used for structural dama-ge detection to indicate the hea...Damage detection in structures is performed via vibra-tion based structural identification. Modal information, such as fre-quencies and mode shapes, are widely used for structural dama-ge detection to indicate the health conditions of civil structures.The deep learning algorithm that works on a multiple layer neuralnetwork model termed as deep autoencoder is proposed to learnthe relationship between the modal information and structural stiff-ness parameters. This is achieved via dimension reduction of themodal information feature and a non-linear regression against thestructural stiffness parameters. Numerical tests on a symmetri-cal steel frame model are conducted to generate the data for thetraining and validation, and to demonstrate the efficiency of theproposed approach for vibration based structural damage detec-tion.展开更多
The linear encoding of a quadtree is an efficient way to represent the quadtree. In this paper, an improved linear quadtree, a cell linear quadtree, is proposed, in which its redundant storage is eliminated and the co...The linear encoding of a quadtree is an efficient way to represent the quadtree. In this paper, an improved linear quadtree, a cell linear quadtree, is proposed, in which its redundant storage is eliminated and the concept of a cell is introduced. Therefore, it has higher storage efficiency than a conventional linear quadtree.展开更多
针对印刷电路板(printed circuit board,PCB)缺陷检测过程中,因包含丰富的小目标缺陷,易出现漏检、误检现象,提出一种基于改进增强金字塔实时检测变换器(enhance pyramid real time detection transformer,EP-RTDETR)小目标PCB表面缺陷...针对印刷电路板(printed circuit board,PCB)缺陷检测过程中,因包含丰富的小目标缺陷,易出现漏检、误检现象,提出一种基于改进增强金字塔实时检测变换器(enhance pyramid real time detection transformer,EP-RTDETR)小目标PCB表面缺陷检测算法。首先,使用CSPDarknet替代原始骨干网络,以增强模型的特征提取能力;其次,设空间移动卷积门控线性单元(spatial moving point convolutional gated linear unit,SMPCGLU)模块改造C2f中的Bottleneck,增强了特征的门控调制能力和空间自适应性;再次,引入可学习位置编码,改进尺度交互机制,增强对不同位置信息的响应能力;然后,基于跨尺度特征融合模块(cross-scale feature-fusion module,CCFM)模块设计小目标增强金字塔结构(small object enhance pyramid,SOEP),增强的特征层和精细的特征融合使模型能够更准确地定位和识别小目标;最后,设计MPDIoU(minimum point distance-based intersection over union)+NWD(normalized wasserstein distance)loss,在加快模型收敛速度的同时更加关注小目标缺陷,回归结果更加准确。试验结果表明,相较于基准模型,准确率P提高了4.6%,召回率R提高了5.1%,平均精度均值mAP50提高了4.6%,参数量减少了16.38 M,浮点数减少了48.3,FPS提高了8.51 s,能够更好地进行小目标PCB表面缺陷检测。展开更多
图神经网络是一种强大的学习图数据的模型,通过节点信息嵌入和图卷积运算实现图结构数据的表示。图数据中节点的结构信息和节点的位置信息对获取图特征至关重要,但现有的图神经网络同时捕获位置信息和结构信息的表达能力有限。对此,提...图神经网络是一种强大的学习图数据的模型,通过节点信息嵌入和图卷积运算实现图结构数据的表示。图数据中节点的结构信息和节点的位置信息对获取图特征至关重要,但现有的图神经网络同时捕获位置信息和结构信息的表达能力有限。对此,提出了一种新的图神经网络——融合位置和结构信息的图神经网络(Positional and Structural Information with Graph Neural Networks, PSI-GNN)。PSI-GNN的核心思想在于利用编码器获取节点的位置和结构信息,并将这些信息特征嵌入到网络中。通过在网络中更新和传递这两种信息,PSI-GNN实现了对位置和结构信息的有效融合与利用,为解决上述问题提供了有效的解决方案。同时,为应对不同类型的图学习任务,PSI-GNN给予位置和结构信息以不同的权重来应对不同的下游任务。为了验证PSI-GNN的有效性,在多个基准图数据集上进行了实验。实验结果表明,PSI-GNN在节点级任务上最高提升了约14%,在图级任务上最高提升了约35%,验证了PSI-GNN在同时捕获位置和结构信息方面的有效性。展开更多
基金Projects(42477162,52108347,52178371,52168046,52178321,52308383)supported by the National Natural Science Foundation of ChinaProjects(2023C03143,2022C01099,2024C01219,2022C03151)supported by the Zhejiang Key Research and Development Plan,China+6 种基金Project(LQ22E080010)supported by the Exploring Youth Project of Zhejiang Natural Science Foundation,ChinaProject(LR21E080005)supported by the Outstanding Youth Project of Natural Science Foundation of Zhejiang Province,ChinaProject(2022M712964)supported by the Postdoctoral Science Foundation of ChinaProject(2023AFB008)supported by the Natural Science Foundation of Hubei Province for Youth,ChinaProject(202203)supported by Engineering Research Centre of Rock-Soil Drilling&Excavation and Protection,Ministry of Education,ChinaProject(202305-2)supported by the Science and Technology Project of Zhejiang Provincial Communication Department,ChinaProject(2021K256)supported by the Construction Research Founds of Department of Housing and Urban-Rural Development of Zhejiang Province,China。
文摘This study proposes a general imperfect thermal contact model to predict the thermal contact resistance at the interface among multi-layered composite structures.Based on the Green-Lindsay(GL)thermoelastic theory,semi analytical solutions of temperature increment and displacement of multi-layered composite structures are obtained by using the Laplace transform method,upon which the effects of thermal resistance coefficient,partition coefficient,thermal conductivity ratio and heat capacity ratio on the responses are studied.The results show that the generalized imperfect thermal contact model can realistically describe the imperfect thermal contact problem.Accordingly,it may degenerate into other thermal contact models by adjusting the thermal resistance coefficient and partition coefficient.
文摘Damage detection in structures is performed via vibra-tion based structural identification. Modal information, such as fre-quencies and mode shapes, are widely used for structural dama-ge detection to indicate the health conditions of civil structures.The deep learning algorithm that works on a multiple layer neuralnetwork model termed as deep autoencoder is proposed to learnthe relationship between the modal information and structural stiff-ness parameters. This is achieved via dimension reduction of themodal information feature and a non-linear regression against thestructural stiffness parameters. Numerical tests on a symmetri-cal steel frame model are conducted to generate the data for thetraining and validation, and to demonstrate the efficiency of theproposed approach for vibration based structural damage detec-tion.
文摘The linear encoding of a quadtree is an efficient way to represent the quadtree. In this paper, an improved linear quadtree, a cell linear quadtree, is proposed, in which its redundant storage is eliminated and the concept of a cell is introduced. Therefore, it has higher storage efficiency than a conventional linear quadtree.
文摘针对印刷电路板(printed circuit board,PCB)缺陷检测过程中,因包含丰富的小目标缺陷,易出现漏检、误检现象,提出一种基于改进增强金字塔实时检测变换器(enhance pyramid real time detection transformer,EP-RTDETR)小目标PCB表面缺陷检测算法。首先,使用CSPDarknet替代原始骨干网络,以增强模型的特征提取能力;其次,设空间移动卷积门控线性单元(spatial moving point convolutional gated linear unit,SMPCGLU)模块改造C2f中的Bottleneck,增强了特征的门控调制能力和空间自适应性;再次,引入可学习位置编码,改进尺度交互机制,增强对不同位置信息的响应能力;然后,基于跨尺度特征融合模块(cross-scale feature-fusion module,CCFM)模块设计小目标增强金字塔结构(small object enhance pyramid,SOEP),增强的特征层和精细的特征融合使模型能够更准确地定位和识别小目标;最后,设计MPDIoU(minimum point distance-based intersection over union)+NWD(normalized wasserstein distance)loss,在加快模型收敛速度的同时更加关注小目标缺陷,回归结果更加准确。试验结果表明,相较于基准模型,准确率P提高了4.6%,召回率R提高了5.1%,平均精度均值mAP50提高了4.6%,参数量减少了16.38 M,浮点数减少了48.3,FPS提高了8.51 s,能够更好地进行小目标PCB表面缺陷检测。
文摘图神经网络是一种强大的学习图数据的模型,通过节点信息嵌入和图卷积运算实现图结构数据的表示。图数据中节点的结构信息和节点的位置信息对获取图特征至关重要,但现有的图神经网络同时捕获位置信息和结构信息的表达能力有限。对此,提出了一种新的图神经网络——融合位置和结构信息的图神经网络(Positional and Structural Information with Graph Neural Networks, PSI-GNN)。PSI-GNN的核心思想在于利用编码器获取节点的位置和结构信息,并将这些信息特征嵌入到网络中。通过在网络中更新和传递这两种信息,PSI-GNN实现了对位置和结构信息的有效融合与利用,为解决上述问题提供了有效的解决方案。同时,为应对不同类型的图学习任务,PSI-GNN给予位置和结构信息以不同的权重来应对不同的下游任务。为了验证PSI-GNN的有效性,在多个基准图数据集上进行了实验。实验结果表明,PSI-GNN在节点级任务上最高提升了约14%,在图级任务上最高提升了约35%,验证了PSI-GNN在同时捕获位置和结构信息方面的有效性。
