Semi-deep foundations socketed in rocks are considered to be a viable option for the foundations in the presence of heavy load imposed by high-rise structures, due to the low settlement and high bearing capacity. In t...Semi-deep foundations socketed in rocks are considered to be a viable option for the foundations in the presence of heavy load imposed by high-rise structures, due to the low settlement and high bearing capacity. In the optimum design of semi-deep foundations, prediction of the shaft bearing capacity, rs, of foundations socketed in rocks is thus critically important. In this study, the unconfined compressive strength(UCS), qu, has been applied in order to investigate the shaft bearing capacity. For this, a database of 106 full-scale load tests is compiled with UCS values of surrounding rocks, in which 34 tests with rock quality designation(RQD), and 5 tests with rock mass rating(RMR). The bearing rocks for semi-deep foundations include limestone, mudstone, siltstone, shale, granite, tuff, granodiorite, claystone, sandstone, phyllite, schist, and greywacke. Using the database, the applicability and accuracy of the existing empirical methods are evaluated and new relations are derived between the shaft bearing capacity and UCS based on the types of rocks. Moreover, a general equation in case of unknown rock types is proposed and it is verified by another set of data. Since rock-socketed shafts are supported by rock mass(not intact rock), a reduction factor for the compressive strength is suggested and verified in which the effect of discontinuities is considered using the modified UCS, qu(modified), based upon RMR and RQD in order to take into account the effect of the rock mass properties.展开更多
Semi-deep foundations socketed in rocks are considered to be a viable option for the foundations in the presence of heavy loads imposed by high-rise buildings and special structures, due to the low settlement and high...Semi-deep foundations socketed in rocks are considered to be a viable option for the foundations in the presence of heavy loads imposed by high-rise buildings and special structures, due to the low settlement and high bearing capacity. In this study, the unconfined compressive strength（UCS） and rock mass cuttability index（RMCI） have been applied to investigating the shaft bearing capacity. For this purpose, a comprehensive database of 178 full-scale load tests is compiled by adding a data set（n = 72）collected by Arioglu et al.（2007） to the data set（n = 106） presented in Rezazadeh and Eslami（2017）.Using the database, the applicability and accuracy of the existing empirical methods are evaluated and new relations are derived between the shaft bearing capacity and UCS/RMCI. Moreover, a general equation in case of unknown rock types is proposed and it is verified by another set of data（series 3 in Rezazadeh and Eslami（2017））. Since rock-socketed shafts are supported by rock mass（not intact rock）,a reduction factor for the compressive strength is suggested and verified in which the effect of discontinuities is considered using the modified UCS, based upon RMR and RQD to consider the effect of the rock mass properties.展开更多
Semi-deep foundations are a remarkable solution in conditions where the soil beneath the foundation is loose to a great depth and there is no possible way to use any way of soil improvement and applying piles would no...Semi-deep foundations are a remarkable solution in conditions where the soil beneath the foundation is loose to a great depth and there is no possible way to use any way of soil improvement and applying piles would not be a logical way considering their cost and time of enforcing. Skirted foundations are a type of semi-deep foundations that can penetrate to the soil up to two times of their breadth. Estimating bearing capacity of these foundations is a long geotechnical problem for engineers whether under absolute or combined loading because of their usage in offshore and onshore projects. For estimating the vertical bearing capacity of these foundations, series of finite element analyses were performed for a range of embedment ratios to investigate the effect of the length of the skirt. The foundation has been modelled with two different types of soil and the results validated with previous analytical, numerical and experimental researches. In addition, the bearing capacity of a skirted foundation under combined loading in V-H space has been analyzed by this approach and the 2-dimentional failure envelope has been presented.展开更多
针对细粒度鸟类检测的数据标注成本高,以及湿地地区鸟类种类繁多、现实场景复杂化等引起的湿地鸟类检测精度低的问题,该研究提出一种基于半监督CST的湿地场景下的细粒度鸟类检测算法(semi-supervised bird detection with CNN and swin ...针对细粒度鸟类检测的数据标注成本高,以及湿地地区鸟类种类繁多、现实场景复杂化等引起的湿地鸟类检测精度低的问题,该研究提出一种基于半监督CST的湿地场景下的细粒度鸟类检测算法(semi-supervised bird detection with CNN and swin transformer,SSBY-CST),首先基于北京14处监测站在不同湿地场景下采集到的图像,构建了涵盖17种鸟类图像数据集,为模型鲁棒性提供可靠数据支撑。其次提出基于伪标签学习法的单阶段半监督学习框架,基于Yolov5主干网络构建教师学生模型,高效利用无标签数据提升检测性能;训练阶段使用双阈值伪标签分配策略替代传统单一阈值伪标签分配,以优化无监督损失函数。然后设计了结合CNN和Swin Transformer的双通道卷积模块CST,以提高不同类别鸟类与湿地背景的区分能力。试验结果表明,仅在100张标注图像下,该文SSBY-CST算法对17种复杂环境下鸟类的检测精准率和mAP@0.5分别为77.5%和58.2%,相比同时期较先进的YOLO模型提升了17.4个百分点和15.5个百分点,在少量标注的前提下实现了较高的检测性能提升,其中黑鹳、西伯利亚银鸥的m AP@0.5分别达到了95.7%和94.5%,相比基线提升了24.9个百分点和14.3个百分点。此外,消融试验分析了双阈值伪标签分配的作用及CST模块的效果,验证了双阈值伪标签分配与CST模块设计的有效性。该框架利用无标注样本在极少量标注量下提升复杂环境下细粒度鸟类检测性能,以加强农林生态的智能数字化管理。该文将半监督扩展到细粒度鸟类检测,为处理农林生态环境下的鸟类检测提供了技术路径。展开更多
针对变工况条件下滚动轴承故障诊断模型泛化性能不佳的问题,基于深度域自适应与半监督学习技术,提出一种带有辅助分类器的半监督卷积神经网络(semi-supervised convolutional neural network based on auxiliary classifier,简称SSCNN-...针对变工况条件下滚动轴承故障诊断模型泛化性能不佳的问题,基于深度域自适应与半监督学习技术,提出一种带有辅助分类器的半监督卷积神经网络(semi-supervised convolutional neural network based on auxiliary classifier,简称SSCNN-AC)滚动轴承跨域故障诊断模型。首先,为提升训练过程中目标域样本伪标签的置信度,所提模型引入最近邻中心分类器作为辅助分类器,以类中心与样本嵌入特征间的余弦距离为目标域样本生成伪标签,有效提升伪标签的可靠性;其次,采用带有标签平滑项的交叉熵损失函数计算分类损失,抑制伪标签噪声对半监督学习的不利影响,提升模型泛化性能;最后,以2个不同数据集的实验结果分析对所提模型进行验证。结果表明:所提模型可有效对齐不同工况下振动信号的嵌入特征,在滚动轴承的跨域故障诊断方面具有明显优势。展开更多
地震相识别是地震数据解释的重要环节之一,深度学习技术可有效提高地震相自动识别的效率和准确性。然而,深度学习方法依赖大量的地震标注数据,在实际应用中标注成本高、难度大,且基础的测井数据无法直接使用。为此,提出了一种基于超稀...地震相识别是地震数据解释的重要环节之一,深度学习技术可有效提高地震相自动识别的效率和准确性。然而,深度学习方法依赖大量的地震标注数据,在实际应用中标注成本高、难度大,且基础的测井数据无法直接使用。为此,提出了一种基于超稀疏测井标注的半监督地震相自动识别方法。首先,在HRNet网络的基础上,构建一种使用一维测井标签进行监督的地震相识别网络模型。其次,针对地震数据的纵向特征,构建稀疏标签采样模块(SLSM)并围绕测井标签采样,不在纵向上对地震数据进行切割,保留其纵向深度特征,为后续的半监督学习任务奠定坚实的基础;最后,针对地震数据的横向相关性,提出区域生长训练策略(RGTS),通过迭代生长的方式将测井标签信息扩展到整个地震体。真实数据实验结果表明,所提出的网络模型仅使用占总数据量不足0.5%的32条一维测井标签,即可实现MIoU(Mean Intersection over Union)为79.64%的地震相识别结果。该方法可为测井资料少且局部分布的工区开展地震相识别研究提供参考,具有良好的应用前景。展开更多
文摘Semi-deep foundations socketed in rocks are considered to be a viable option for the foundations in the presence of heavy load imposed by high-rise structures, due to the low settlement and high bearing capacity. In the optimum design of semi-deep foundations, prediction of the shaft bearing capacity, rs, of foundations socketed in rocks is thus critically important. In this study, the unconfined compressive strength(UCS), qu, has been applied in order to investigate the shaft bearing capacity. For this, a database of 106 full-scale load tests is compiled with UCS values of surrounding rocks, in which 34 tests with rock quality designation(RQD), and 5 tests with rock mass rating(RMR). The bearing rocks for semi-deep foundations include limestone, mudstone, siltstone, shale, granite, tuff, granodiorite, claystone, sandstone, phyllite, schist, and greywacke. Using the database, the applicability and accuracy of the existing empirical methods are evaluated and new relations are derived between the shaft bearing capacity and UCS based on the types of rocks. Moreover, a general equation in case of unknown rock types is proposed and it is verified by another set of data. Since rock-socketed shafts are supported by rock mass(not intact rock), a reduction factor for the compressive strength is suggested and verified in which the effect of discontinuities is considered using the modified UCS, qu(modified), based upon RMR and RQD in order to take into account the effect of the rock mass properties.
文摘Semi-deep foundations socketed in rocks are considered to be a viable option for the foundations in the presence of heavy loads imposed by high-rise buildings and special structures, due to the low settlement and high bearing capacity. In this study, the unconfined compressive strength（UCS） and rock mass cuttability index（RMCI） have been applied to investigating the shaft bearing capacity. For this purpose, a comprehensive database of 178 full-scale load tests is compiled by adding a data set（n = 72）collected by Arioglu et al.（2007） to the data set（n = 106） presented in Rezazadeh and Eslami（2017）.Using the database, the applicability and accuracy of the existing empirical methods are evaluated and new relations are derived between the shaft bearing capacity and UCS/RMCI. Moreover, a general equation in case of unknown rock types is proposed and it is verified by another set of data（series 3 in Rezazadeh and Eslami（2017））. Since rock-socketed shafts are supported by rock mass（not intact rock）,a reduction factor for the compressive strength is suggested and verified in which the effect of discontinuities is considered using the modified UCS, based upon RMR and RQD to consider the effect of the rock mass properties.
文摘Semi-deep foundations are a remarkable solution in conditions where the soil beneath the foundation is loose to a great depth and there is no possible way to use any way of soil improvement and applying piles would not be a logical way considering their cost and time of enforcing. Skirted foundations are a type of semi-deep foundations that can penetrate to the soil up to two times of their breadth. Estimating bearing capacity of these foundations is a long geotechnical problem for engineers whether under absolute or combined loading because of their usage in offshore and onshore projects. For estimating the vertical bearing capacity of these foundations, series of finite element analyses were performed for a range of embedment ratios to investigate the effect of the length of the skirt. The foundation has been modelled with two different types of soil and the results validated with previous analytical, numerical and experimental researches. In addition, the bearing capacity of a skirted foundation under combined loading in V-H space has been analyzed by this approach and the 2-dimentional failure envelope has been presented.
文摘针对细粒度鸟类检测的数据标注成本高,以及湿地地区鸟类种类繁多、现实场景复杂化等引起的湿地鸟类检测精度低的问题,该研究提出一种基于半监督CST的湿地场景下的细粒度鸟类检测算法(semi-supervised bird detection with CNN and swin transformer,SSBY-CST),首先基于北京14处监测站在不同湿地场景下采集到的图像,构建了涵盖17种鸟类图像数据集,为模型鲁棒性提供可靠数据支撑。其次提出基于伪标签学习法的单阶段半监督学习框架,基于Yolov5主干网络构建教师学生模型,高效利用无标签数据提升检测性能;训练阶段使用双阈值伪标签分配策略替代传统单一阈值伪标签分配,以优化无监督损失函数。然后设计了结合CNN和Swin Transformer的双通道卷积模块CST,以提高不同类别鸟类与湿地背景的区分能力。试验结果表明,仅在100张标注图像下,该文SSBY-CST算法对17种复杂环境下鸟类的检测精准率和mAP@0.5分别为77.5%和58.2%,相比同时期较先进的YOLO模型提升了17.4个百分点和15.5个百分点,在少量标注的前提下实现了较高的检测性能提升,其中黑鹳、西伯利亚银鸥的m AP@0.5分别达到了95.7%和94.5%,相比基线提升了24.9个百分点和14.3个百分点。此外,消融试验分析了双阈值伪标签分配的作用及CST模块的效果,验证了双阈值伪标签分配与CST模块设计的有效性。该框架利用无标注样本在极少量标注量下提升复杂环境下细粒度鸟类检测性能,以加强农林生态的智能数字化管理。该文将半监督扩展到细粒度鸟类检测,为处理农林生态环境下的鸟类检测提供了技术路径。
文摘针对变工况条件下滚动轴承故障诊断模型泛化性能不佳的问题,基于深度域自适应与半监督学习技术,提出一种带有辅助分类器的半监督卷积神经网络(semi-supervised convolutional neural network based on auxiliary classifier,简称SSCNN-AC)滚动轴承跨域故障诊断模型。首先,为提升训练过程中目标域样本伪标签的置信度,所提模型引入最近邻中心分类器作为辅助分类器,以类中心与样本嵌入特征间的余弦距离为目标域样本生成伪标签,有效提升伪标签的可靠性;其次,采用带有标签平滑项的交叉熵损失函数计算分类损失,抑制伪标签噪声对半监督学习的不利影响,提升模型泛化性能;最后,以2个不同数据集的实验结果分析对所提模型进行验证。结果表明:所提模型可有效对齐不同工况下振动信号的嵌入特征,在滚动轴承的跨域故障诊断方面具有明显优势。
文摘地震相识别是地震数据解释的重要环节之一,深度学习技术可有效提高地震相自动识别的效率和准确性。然而,深度学习方法依赖大量的地震标注数据,在实际应用中标注成本高、难度大,且基础的测井数据无法直接使用。为此,提出了一种基于超稀疏测井标注的半监督地震相自动识别方法。首先,在HRNet网络的基础上,构建一种使用一维测井标签进行监督的地震相识别网络模型。其次,针对地震数据的纵向特征,构建稀疏标签采样模块(SLSM)并围绕测井标签采样,不在纵向上对地震数据进行切割,保留其纵向深度特征,为后续的半监督学习任务奠定坚实的基础;最后,针对地震数据的横向相关性,提出区域生长训练策略(RGTS),通过迭代生长的方式将测井标签信息扩展到整个地震体。真实数据实验结果表明,所提出的网络模型仅使用占总数据量不足0.5%的32条一维测井标签,即可实现MIoU(Mean Intersection over Union)为79.64%的地震相识别结果。该方法可为测井资料少且局部分布的工区开展地震相识别研究提供参考,具有良好的应用前景。
