A three-dimensional finite element model was established for the milling of thin-walled parts. The physical model of the milling of the part was established using the AdvantEdge FEM software as the platform. The alumi...A three-dimensional finite element model was established for the milling of thin-walled parts. The physical model of the milling of the part was established using the AdvantEdge FEM software as the platform. The aluminum alloy impeller was designated as the object to be processed and the boundary conditions which met the actual machining were set. Through the solution, the physical quantities such as the three-way cutting force, the tool temperature, and the tool stress were obtained, and the calculation of the elastic deformation of the thin-walled blade of the free-form surface at the contact points between the tool and the workpiece was realized. The elastic deformation law of the thin-walled blade was then predicted. The results show that the maximum deviation between the predicted value and the actual measured machining value of the elastic deformation was 26.055 μm; the minimum deviation was 2.011 μm, with the average deviation being 10.154 μm. This shows that the prediction is in close agreement with the actual result.展开更多
针对行人检测系统中存在的难以同时具有较高的检测率和较快的检测速度这一问题,本文提出了一种自适应Coarse-to-Fine Deformable Part Model(CtF DPM)的行人检测模型.首先,将低分辨率根滤波器特征提取得分与阈值进行比较,以确定高分辨...针对行人检测系统中存在的难以同时具有较高的检测率和较快的检测速度这一问题,本文提出了一种自适应Coarse-to-Fine Deformable Part Model(CtF DPM)的行人检测模型.首先,将低分辨率根滤波器特征提取得分与阈值进行比较,以确定高分辨率部件滤波器的特征提取区域;随后,在同分辨率层中引入同级约束关系,增强同层的特征相关性;最后,将该特征提取与其他多种算法在INRIA数据库中进行检测准确性测试,并与隐式支持向量机(LSVM)结合进行实际道路环境测试.理论性能和实际测试结果表明:基于自适应CtF DPM的行人检测模型在保证检测性能的同时,特征提取时间可降至十几毫秒,显著提高了检测速度.展开更多
针对可变形部件模型(deformable parts model,DPM)同等对待各部件,无法体现不同部件对识别过程的贡献度差异的不足,提出一种权重系数可变形模型(weighted coefficient deformable parts model,WCDPM),对DPM中的各部件赋予权重,强调区分...针对可变形部件模型(deformable parts model,DPM)同等对待各部件,无法体现不同部件对识别过程的贡献度差异的不足,提出一种权重系数可变形模型(weighted coefficient deformable parts model,WCDPM),对DPM中的各部件赋予权重,强调区分度较高的部件在识别过程的作用,弱化区分度低的部件对识别的影响,提高细粒度识别精度.同时给出了模型的训练过程和权重系数的学习方法.在Airplan OID和Oxford-IIIT Pet两个数据集上进行实验,验证了该方法的有效性.展开更多
行人检测是计算机视觉技术中一个热门的研究热点,在汽车辅助驾驶和视频监控等方面具有重要作用.由于传统的可变形部件模型(deformable part model,DPM)采用滑动窗口检测方式,在背景区域花费大量检测时间会导致检测速度降低,因此提出了...行人检测是计算机视觉技术中一个热门的研究热点,在汽车辅助驾驶和视频监控等方面具有重要作用.由于传统的可变形部件模型(deformable part model,DPM)采用滑动窗口检测方式,在背景区域花费大量检测时间会导致检测速度降低,因此提出了一种基于BING-casDPM的快速行人检测算法.首先基于二进制化梯度范数特征(binarized normed gradient,BING)训练一个二级支持向量机(support vector machine,SVM)分类器,通过该分类器快速标定出所测图像中包含各类物体的候选区域;然后根据候选区域窗口的特点进一步提取待检测框;最后将待检测框作为输入,使用级联DPM(cascade DPM,casDPM)模型进行精确检测,并将结果返回至原图.实验结果表明,该算法在基本不降低检测率的情况下,其检测速度比经典DPM模型检测速度提高了约16倍,比casDPM模型提高了约40%.展开更多
In the process of thin-wall parts assembly for an antenna,the parts assembly deformation deviation is occurring due to the riveting assembly.In view of the riveting assembly deformation problems,it can be analyzed thr...In the process of thin-wall parts assembly for an antenna,the parts assembly deformation deviation is occurring due to the riveting assembly.In view of the riveting assembly deformation problems,it can be analyzed through transient and static simulation.In this work,the theoretical deformation model for riveting assembly is established with round head rivet.The simulation analysis for riveting deformation is carried out with the riveting assembly piece including four rivets,which comparing with the measuring points experiment results of riveting test piece through dealing with the experimental data using the point coordinate transform method and the space line fitting method.Simultaneously,the deformation deviation of the overall thin-wall parts assembly structure is analyzed through finite element simulation;and its results are verified by the measuring experiment for riveting assembly with the deformation deviation of some key points on the thin-wall parts.Through the comparison analysis,it is shown that the simulation results agree well with the experimental results,which proves the correctness and effectiveness of the theoretical analysis,simulation results and the given experiment data processing method.Through the study on the riveting assembly for thin-wall parts,it will provide a theoretical foundation for improving thin-wall parts assembly quality of large antenna in future.展开更多
提出了基于可变形部件模型(deformable part model,DPM)的高分二号(GaoFen-2,GF2)遥感影像船只检测方法,并与区域卷积网络(regional convolutional neural network,R-CNN)进行比较。先将遥感影像分段以获得船只的粗略感兴趣区域(regions...提出了基于可变形部件模型(deformable part model,DPM)的高分二号(GaoFen-2,GF2)遥感影像船只检测方法,并与区域卷积网络(regional convolutional neural network,R-CNN)进行比较。先将遥感影像分段以获得船只的粗略感兴趣区域(regions of interest,ROI),然后在ROI内计算方向梯度直方图(histogram of oriented gradients,HOG)和卷积特征,再分别由DPM和R-CNN采用HOG和卷积特征。为测试R-CNN的最佳性能,将具有5个卷积层(ZF网)和具有13个卷积层(VGG网)的网络应用于船只检测。使用8张GF2遥感影像的3 523艘船只的实验结果表明,DPM和R-CNN都能以高召回率和正确率检测水中的船只,但对于聚集船只而言,DPM的效果优于R-CNN。基于HOG+DPM,ZF网和VGG网的方法平均精度分别为95.031%,93.282%和93.683%。展开更多
基金Project(U1530138)supported by the National Natural Science Foundation of ChinaProject(A1-8903-17-0103)supported by the Natural Science Foundation of Shanghai Municipal Education Commission,China
文摘A three-dimensional finite element model was established for the milling of thin-walled parts. The physical model of the milling of the part was established using the AdvantEdge FEM software as the platform. The aluminum alloy impeller was designated as the object to be processed and the boundary conditions which met the actual machining were set. Through the solution, the physical quantities such as the three-way cutting force, the tool temperature, and the tool stress were obtained, and the calculation of the elastic deformation of the thin-walled blade of the free-form surface at the contact points between the tool and the workpiece was realized. The elastic deformation law of the thin-walled blade was then predicted. The results show that the maximum deviation between the predicted value and the actual measured machining value of the elastic deformation was 26.055 μm; the minimum deviation was 2.011 μm, with the average deviation being 10.154 μm. This shows that the prediction is in close agreement with the actual result.
文摘针对可变形部件模型(deformable parts model,DPM)同等对待各部件,无法体现不同部件对识别过程的贡献度差异的不足,提出一种权重系数可变形模型(weighted coefficient deformable parts model,WCDPM),对DPM中的各部件赋予权重,强调区分度较高的部件在识别过程的作用,弱化区分度低的部件对识别的影响,提高细粒度识别精度.同时给出了模型的训练过程和权重系数的学习方法.在Airplan OID和Oxford-IIIT Pet两个数据集上进行实验,验证了该方法的有效性.
基金Project(51675100)supported by the National Natural Science Foundation of ChinaProject(2016ZX04004008)supported by the National Numerical Control Equipment Major Project of ChinaProject(6902002116)supported by the Foundation of Certain Ministry of China
文摘In the process of thin-wall parts assembly for an antenna,the parts assembly deformation deviation is occurring due to the riveting assembly.In view of the riveting assembly deformation problems,it can be analyzed through transient and static simulation.In this work,the theoretical deformation model for riveting assembly is established with round head rivet.The simulation analysis for riveting deformation is carried out with the riveting assembly piece including four rivets,which comparing with the measuring points experiment results of riveting test piece through dealing with the experimental data using the point coordinate transform method and the space line fitting method.Simultaneously,the deformation deviation of the overall thin-wall parts assembly structure is analyzed through finite element simulation;and its results are verified by the measuring experiment for riveting assembly with the deformation deviation of some key points on the thin-wall parts.Through the comparison analysis,it is shown that the simulation results agree well with the experimental results,which proves the correctness and effectiveness of the theoretical analysis,simulation results and the given experiment data processing method.Through the study on the riveting assembly for thin-wall parts,it will provide a theoretical foundation for improving thin-wall parts assembly quality of large antenna in future.
文摘提出了基于可变形部件模型(deformable part model,DPM)的高分二号(GaoFen-2,GF2)遥感影像船只检测方法,并与区域卷积网络(regional convolutional neural network,R-CNN)进行比较。先将遥感影像分段以获得船只的粗略感兴趣区域(regions of interest,ROI),然后在ROI内计算方向梯度直方图(histogram of oriented gradients,HOG)和卷积特征,再分别由DPM和R-CNN采用HOG和卷积特征。为测试R-CNN的最佳性能,将具有5个卷积层(ZF网)和具有13个卷积层(VGG网)的网络应用于船只检测。使用8张GF2遥感影像的3 523艘船只的实验结果表明,DPM和R-CNN都能以高召回率和正确率检测水中的船只,但对于聚集船只而言,DPM的效果优于R-CNN。基于HOG+DPM,ZF网和VGG网的方法平均精度分别为95.031%,93.282%和93.683%。
