针对目前基于信道脉冲响应(Channel Impulse Response,CIR)的非视距(None Line of Sight,NLoS)/视距(Line of Sight,LoS)识别方法精度低、泛化能力差的问题,提出了一种多层卷积神经网络(Convolutional Neural Network,CNN)与通道注意力...针对目前基于信道脉冲响应(Channel Impulse Response,CIR)的非视距(None Line of Sight,NLoS)/视距(Line of Sight,LoS)识别方法精度低、泛化能力差的问题,提出了一种多层卷积神经网络(Convolutional Neural Network,CNN)与通道注意力模块(Channel Attention Module,CAM)相结合的NLoS/LoS识别方法。在多层CNN中嵌入CAM提取原始CIR的时域数据特征,利用全局平均池化层代替全连接层进行特征整合并分类输出。使用欧洲地平线2020计划项目eWINE公开的数据集进行不同结构模型和不同识别方法的对比实验,结果表明,所提出的CNN-CAM模型LoS和NLoS召回率分别达到了92.29%与87.71%,准确率达到了90.00%,F1分数达到了90.22%。与现有多种传统识别方法相比,均具有更好的识别效果。展开更多
UFLD(ultra fast structure aware deep lane detection)是一种轻量化车道线检测模型,为提升模型的检测精度,对模型进行改进。引入CAM(channel attention mechanism)使模型能更关注携带重要车道线信息的特征通道和像素;为了感知车道线...UFLD(ultra fast structure aware deep lane detection)是一种轻量化车道线检测模型,为提升模型的检测精度,对模型进行改进。引入CAM(channel attention mechanism)使模型能更关注携带重要车道线信息的特征通道和像素;为了感知车道线的细节信息,引入ASPP(atrous spatial pyramid pooling)扩大卷积过程的感受野,提高模型分割精度;搭建引入CAM和ASPP后的改进模型,并在改进的模型上进行实验。实验结果表明:在TuSimple数据集上以ResNet18为主干网络的模型检测精度由95.81%提升至95.98%,以ResNet34为主干网络的模型检测精度由95.84%提升至96.12%;在CULane数据集上,无论是以ResNet18还是以ResNet34为主干网络模型,其平均精度均有不同程度的提高。展开更多
An elastoplastic constitutive model based on the Modified Cam Clay(MCC)model is developed to describe the mechanical behaviour of soils cemented via microbially induced calcite precipitation(MICP).It considers the inc...An elastoplastic constitutive model based on the Modified Cam Clay(MCC)model is developed to describe the mechanical behaviour of soils cemented via microbially induced calcite precipitation(MICP).It considers the increase of the elastic stiffness,the change of the yield surface due to MICP cementation and the degradation of calcium carbonate bonds during shearing.Specifically,to capture the typical contraction-dilation transition in MICP soils,the original volumetric hardening rule in the MCC model is modified to a combined deviatoric and volumetric hardening rule.The model could reproduce a series of drained triaxial tests on MICP-treated soils with different calcium carbonate contents.Further,we carry out a parametric study and observe numerical instability in some cases.In combination with an analytical analysis,our numerical modelling has identified the benefits and limitations of using MCCbased models in the simulation of MICP-cemented soils,leading to suggestions for further model development.展开更多
文摘针对目前基于信道脉冲响应(Channel Impulse Response,CIR)的非视距(None Line of Sight,NLoS)/视距(Line of Sight,LoS)识别方法精度低、泛化能力差的问题,提出了一种多层卷积神经网络(Convolutional Neural Network,CNN)与通道注意力模块(Channel Attention Module,CAM)相结合的NLoS/LoS识别方法。在多层CNN中嵌入CAM提取原始CIR的时域数据特征,利用全局平均池化层代替全连接层进行特征整合并分类输出。使用欧洲地平线2020计划项目eWINE公开的数据集进行不同结构模型和不同识别方法的对比实验,结果表明,所提出的CNN-CAM模型LoS和NLoS召回率分别达到了92.29%与87.71%,准确率达到了90.00%,F1分数达到了90.22%。与现有多种传统识别方法相比,均具有更好的识别效果。
文摘UFLD(ultra fast structure aware deep lane detection)是一种轻量化车道线检测模型,为提升模型的检测精度,对模型进行改进。引入CAM(channel attention mechanism)使模型能更关注携带重要车道线信息的特征通道和像素;为了感知车道线的细节信息,引入ASPP(atrous spatial pyramid pooling)扩大卷积过程的感受野,提高模型分割精度;搭建引入CAM和ASPP后的改进模型,并在改进的模型上进行实验。实验结果表明:在TuSimple数据集上以ResNet18为主干网络的模型检测精度由95.81%提升至95.98%,以ResNet34为主干网络的模型检测精度由95.84%提升至96.12%;在CULane数据集上,无论是以ResNet18还是以ResNet34为主干网络模型,其平均精度均有不同程度的提高。
基金funded by the German Research Foundation(DFG)(Grant No.NA 330/20e1).
文摘An elastoplastic constitutive model based on the Modified Cam Clay(MCC)model is developed to describe the mechanical behaviour of soils cemented via microbially induced calcite precipitation(MICP).It considers the increase of the elastic stiffness,the change of the yield surface due to MICP cementation and the degradation of calcium carbonate bonds during shearing.Specifically,to capture the typical contraction-dilation transition in MICP soils,the original volumetric hardening rule in the MCC model is modified to a combined deviatoric and volumetric hardening rule.The model could reproduce a series of drained triaxial tests on MICP-treated soils with different calcium carbonate contents.Further,we carry out a parametric study and observe numerical instability in some cases.In combination with an analytical analysis,our numerical modelling has identified the benefits and limitations of using MCCbased models in the simulation of MICP-cemented soils,leading to suggestions for further model development.
