In an automatic bobbin management system that simultaneously detects bobbin color and residual yarn,a composite texture segmentation and recognition operation based on an odd partial Gabor filter and multi-color space...In an automatic bobbin management system that simultaneously detects bobbin color and residual yarn,a composite texture segmentation and recognition operation based on an odd partial Gabor filter and multi-color space hierarchical clustering are proposed.Firstly,the parameter-optimized odd partial Gabor filter is used to distinguish bobbin and yarn texture,to explore Garbor parameters for yarn bobbins,and to accurately discriminate frequency characteristics of yarns and texture.Secondly,multi-color clustering segmentation using color spaces such as red,green,blue(RGB)and CIELUV(LUV)solves the problems of over-segmentation and segmentation errors,which are caused by the difficulty of accurately representing the complex and variable color information of yarns in a single-color space and the low contrast between the target and background.Finally,the segmented bobbin is combined with the odd partial Gabor’s edge recognition operator to further distinguish bobbin texture from yarn texture and locate the position and size of the residual yarn.Experimental results show that the method is robust in identifying complex texture,damaged and dyed bobbins,and multi-color yarns.Residual yarn identification can distinguish texture features and residual yarns well and it can be transferred to the detection and differentiation of complex texture,which is significantly better than traditional methods.展开更多
High fidelity analysis are utilized in modern engineering design optimization problems which involve expensive black-box models.For computation-intensive engineering design problems,efficient global optimization metho...High fidelity analysis are utilized in modern engineering design optimization problems which involve expensive black-box models.For computation-intensive engineering design problems,efficient global optimization methods must be developed to relieve the computational burden.A new metamodel-based global optimization method using fuzzy clustering for design space reduction(MGO-FCR) is presented.The uniformly distributed initial sample points are generated by Latin hypercube design to construct the radial basis function metamodel,whose accuracy is improved with increasing number of sample points gradually.Fuzzy c-mean method and Gath-Geva clustering method are applied to divide the design space into several small interesting cluster spaces for low and high dimensional problems respectively.Modeling efficiency and accuracy are directly related to the design space,so unconcerned spaces are eliminated by the proposed reduction principle and two pseudo reduction algorithms.The reduction principle is developed to determine whether the current design space should be reduced and which space is eliminated.The first pseudo reduction algorithm improves the speed of clustering,while the second pseudo reduction algorithm ensures the design space to be reduced.Through several numerical benchmark functions,comparative studies with adaptive response surface method,approximated unimodal region elimination method and mode-pursuing sampling are carried out.The optimization results reveal that this method captures the real global optimum for all the numerical benchmark functions.And the number of function evaluations show that the efficiency of this method is favorable especially for high dimensional problems.Based on this global design optimization method,a design optimization of a lifting surface in high speed flow is carried out and this method saves about 10 h compared with genetic algorithms.This method possesses favorable performance on efficiency,robustness and capability of global convergence and gives a new optimization strategy for engineering design optimization problems involving expensive black box models.展开更多
Suppose {X,X n,n≥1} are the i.i.d. random variables with values in 2 type Banach space B,S n=∑nk=1X k,φ(x) is a increasing function on [0,+∞), φ(x)→+∞, and φ(x)x is no increasing; then we point out that th...Suppose {X,X n,n≥1} are the i.i.d. random variables with values in 2 type Banach space B,S n=∑nk=1X k,φ(x) is a increasing function on [0,+∞), φ(x)→+∞, and φ(x)x is no increasing; then we point out that the cluster set CS n2nφ(n) is ρK, where ρ= lim L 2nφ(n)<+∞, ∫ ∞e -ρ 2φ(x) 1xd x=+∞,X∈WM 2 0, and E‖X‖ 2φ(‖X‖)<+∞.展开更多
By using the SLC(Single-Link Cluster)method,this study worked in three respects:(a)set up three-dimensional(3-D)SLC software that can deal with a large catalogue of earthquakes and analyze the characteristics of earth...By using the SLC(Single-Link Cluster)method,this study worked in three respects:(a)set up three-dimensional(3-D)SLC software that can deal with a large catalogue of earthquakes and analyze the characteristics of earthquakes’ clustering and scattering in time-space:(b)defined several parameters to describe the distinguishing feature for the SLC frame and developed a technique to calculate the 3-D SLC frames and these parameters with gradual time-sliding,and inspected their variations with time,especially before large events; and(c)by using these means,treated the earthquake catalogue in the top area of the Kunlun-Altun-Arc as well as some valuable results that had been obtained.展开更多
深度嵌入聚类(deep embedding clustering,DEC)算法只通过自编码器,以单一实例重构的方式将数据嵌入到低维矢量化特征空间中进行聚类,而忽略了不同实例之间的关系,导致可能无法很好地区分嵌入空间中的实例。针对上述问题,提出基于对比...深度嵌入聚类(deep embedding clustering,DEC)算法只通过自编码器,以单一实例重构的方式将数据嵌入到低维矢量化特征空间中进行聚类,而忽略了不同实例之间的关系,导致可能无法很好地区分嵌入空间中的实例。针对上述问题,提出基于对比学习的矢量化特征空间嵌入聚类(vectorized feature space embedded clustering based on contrastive learning,VECCL)方法。通过对比学习以辨识数据实例之间异同性的方式,从数据中提取出具有同近异远聚类语义的特征,并作为先验知识带入DEC中,引导自编码器初始化带有深层数据信息的低维聚类特征空间。同时利用软分类标签构造熵损失,与自编码器的重构损失一起作为正则化项引入聚类损失函数中,共同细化聚类。实验结果表明,所提方法提取特征的能力更强,与DEC方法在数据集CIFAR10、CIFAR100和STL10上的实验结果相比,ACC分别提升48.1个百分点、23.1个百分点和41.8个百分点,NMI分别提升41.0个百分点、25.2个百分点和39.0个百分点,ARI分别提升45.4个百分点、16.4个百分点和41.8个百分点。展开更多
为总结全球范围内煤矿安全管理的研究现状并预测其未来发展趋势,借助Cite Space V软件,对1992--2017年期间Web of Science(wos)收录的“煤矿”“安全管理”领域的相关文献进行可视化知识图谱分析。通过对该领域主要研究机构及国家、作...为总结全球范围内煤矿安全管理的研究现状并预测其未来发展趋势,借助Cite Space V软件,对1992--2017年期间Web of Science(wos)收录的“煤矿”“安全管理”领域的相关文献进行可视化知识图谱分析。通过对该领域主要研究机构及国家、作者等方面进行可视化分析,描述该领域的主要研究力量及研究人员的分布;通过对关键词、研究聚类等进行知识图谱可视化分析,描述该领域的研究热点及未来发展趋势。结果表明:该领域的研究主要集中于中国、美国,且主要研究机构大多为各国高校,作者间合作以小范围团队合作为主;研究主要集中于安全教育、人因失误、数据挖掘以及职业病防治等方面;数据挖掘是该领域未来研究发展趋势之一。展开更多
基金Key Research and Development Plan of Shaanxi Province,China(No.2023-YBGY-330)。
文摘In an automatic bobbin management system that simultaneously detects bobbin color and residual yarn,a composite texture segmentation and recognition operation based on an odd partial Gabor filter and multi-color space hierarchical clustering are proposed.Firstly,the parameter-optimized odd partial Gabor filter is used to distinguish bobbin and yarn texture,to explore Garbor parameters for yarn bobbins,and to accurately discriminate frequency characteristics of yarns and texture.Secondly,multi-color clustering segmentation using color spaces such as red,green,blue(RGB)and CIELUV(LUV)solves the problems of over-segmentation and segmentation errors,which are caused by the difficulty of accurately representing the complex and variable color information of yarns in a single-color space and the low contrast between the target and background.Finally,the segmented bobbin is combined with the odd partial Gabor’s edge recognition operator to further distinguish bobbin texture from yarn texture and locate the position and size of the residual yarn.Experimental results show that the method is robust in identifying complex texture,damaged and dyed bobbins,and multi-color yarns.Residual yarn identification can distinguish texture features and residual yarns well and it can be transferred to the detection and differentiation of complex texture,which is significantly better than traditional methods.
基金supported by National Natural Science Foundation of China(Grant No.51105040)Aeronautic Science Foundation of China(Grant No.2011ZA72003)Excellent Young Scholars Research Fund of Beijing Institute of Technology(Grant No.2010Y0102)
文摘High fidelity analysis are utilized in modern engineering design optimization problems which involve expensive black-box models.For computation-intensive engineering design problems,efficient global optimization methods must be developed to relieve the computational burden.A new metamodel-based global optimization method using fuzzy clustering for design space reduction(MGO-FCR) is presented.The uniformly distributed initial sample points are generated by Latin hypercube design to construct the radial basis function metamodel,whose accuracy is improved with increasing number of sample points gradually.Fuzzy c-mean method and Gath-Geva clustering method are applied to divide the design space into several small interesting cluster spaces for low and high dimensional problems respectively.Modeling efficiency and accuracy are directly related to the design space,so unconcerned spaces are eliminated by the proposed reduction principle and two pseudo reduction algorithms.The reduction principle is developed to determine whether the current design space should be reduced and which space is eliminated.The first pseudo reduction algorithm improves the speed of clustering,while the second pseudo reduction algorithm ensures the design space to be reduced.Through several numerical benchmark functions,comparative studies with adaptive response surface method,approximated unimodal region elimination method and mode-pursuing sampling are carried out.The optimization results reveal that this method captures the real global optimum for all the numerical benchmark functions.And the number of function evaluations show that the efficiency of this method is favorable especially for high dimensional problems.Based on this global design optimization method,a design optimization of a lifting surface in high speed flow is carried out and this method saves about 10 h compared with genetic algorithms.This method possesses favorable performance on efficiency,robustness and capability of global convergence and gives a new optimization strategy for engineering design optimization problems involving expensive black box models.
文摘Suppose {X,X n,n≥1} are the i.i.d. random variables with values in 2 type Banach space B,S n=∑nk=1X k,φ(x) is a increasing function on [0,+∞), φ(x)→+∞, and φ(x)x is no increasing; then we point out that the cluster set CS n2nφ(n) is ρK, where ρ= lim L 2nφ(n)<+∞, ∫ ∞e -ρ 2φ(x) 1xd x=+∞,X∈WM 2 0, and E‖X‖ 2φ(‖X‖)<+∞.
基金This project was sponsored by the United Earthquake Science Foundation (93068), China
文摘By using the SLC(Single-Link Cluster)method,this study worked in three respects:(a)set up three-dimensional(3-D)SLC software that can deal with a large catalogue of earthquakes and analyze the characteristics of earthquakes’ clustering and scattering in time-space:(b)defined several parameters to describe the distinguishing feature for the SLC frame and developed a technique to calculate the 3-D SLC frames and these parameters with gradual time-sliding,and inspected their variations with time,especially before large events; and(c)by using these means,treated the earthquake catalogue in the top area of the Kunlun-Altun-Arc as well as some valuable results that had been obtained.
文摘针对现有的深度获取方式存在数据缺失、分辨率低等问题,提出一种基于软聚类的深度图增强方法,称为软聚类求解器.该方法利用软聚类的强边缘保持特性提高深度图增强的精度.将软聚类仿射矩阵和加权最小二乘模型有机结合,构建了软聚类求解器中的置信加权最小二乘模型,提出了基于迭代的求解方法.为评估所提出的方法,在多项深度图增强任务上进行试验,包括深度图补洞、深度图超分辨率和深度图纠正,评价指标包含了峰值信噪比(PSNR)、结构相似度(SSIM)、均方根差(RMSE)和运行效率.结果表明:文中方法在深度图补洞任务中的平均PSNR达到了42.28,平均SSIM达到了98.83%;在深度图超分辨率、深度图纠正任务中的平均RMSE达到了8.96、 2.36.文中方法处理1张分辨率为2 048×1 024像素的图像仅需5.03 s.
文摘深度嵌入聚类(deep embedding clustering,DEC)算法只通过自编码器,以单一实例重构的方式将数据嵌入到低维矢量化特征空间中进行聚类,而忽略了不同实例之间的关系,导致可能无法很好地区分嵌入空间中的实例。针对上述问题,提出基于对比学习的矢量化特征空间嵌入聚类(vectorized feature space embedded clustering based on contrastive learning,VECCL)方法。通过对比学习以辨识数据实例之间异同性的方式,从数据中提取出具有同近异远聚类语义的特征,并作为先验知识带入DEC中,引导自编码器初始化带有深层数据信息的低维聚类特征空间。同时利用软分类标签构造熵损失,与自编码器的重构损失一起作为正则化项引入聚类损失函数中,共同细化聚类。实验结果表明,所提方法提取特征的能力更强,与DEC方法在数据集CIFAR10、CIFAR100和STL10上的实验结果相比,ACC分别提升48.1个百分点、23.1个百分点和41.8个百分点,NMI分别提升41.0个百分点、25.2个百分点和39.0个百分点,ARI分别提升45.4个百分点、16.4个百分点和41.8个百分点。
文摘为总结全球范围内煤矿安全管理的研究现状并预测其未来发展趋势,借助Cite Space V软件,对1992--2017年期间Web of Science(wos)收录的“煤矿”“安全管理”领域的相关文献进行可视化知识图谱分析。通过对该领域主要研究机构及国家、作者等方面进行可视化分析,描述该领域的主要研究力量及研究人员的分布;通过对关键词、研究聚类等进行知识图谱可视化分析,描述该领域的研究热点及未来发展趋势。结果表明:该领域的研究主要集中于中国、美国,且主要研究机构大多为各国高校,作者间合作以小范围团队合作为主;研究主要集中于安全教育、人因失误、数据挖掘以及职业病防治等方面;数据挖掘是该领域未来研究发展趋势之一。
