改进的Faster-RCNN算法在聚乙烯管接头内部缺陷检测中的应用

超声相控阵技术是目前聚乙烯管道热熔接头内部缺陷检测的一种主流方法。提出了基于注意力机制的改进Faster-RCNN目标检测网络用于超声相控阵D扫图聚乙烯管接头内部缺陷检测。针对聚乙烯管道热熔接头内部超声相控阵D扫图小缺陷较多、特征信息容易丢失的问题,将残差网络(ResNet50)与特征金字塔网络(FPN)相结合作为骨干网络,并引入卷积注意力模块(CBAM)自适应细化特征。将SSD网络框架和Faster-RCNN网络框架用于模型训练和测试,使用VGG16、ResNet50、ResNet50+FPN、ACBM+ResNet50+FPN作为骨干网络依次对超声相控阵聚乙烯管道热熔对接接头内部缺陷样本进行训练对比。结果表明,改进的Faster-RCNN网络模型在聚乙烯管接头内部缺陷检测和分类方面有明显改进,对小缺陷的检测性能有了显著的提高。     

Layer Dependence of Graphene for Oxidation Resistance of Cu Surface

We studied the oxidation resistance of graphene-coated Cu surface and its layer dependence by directly growing monolayer graphene with different multilayer structures coexisted, di-minishing the influence induced by residue and transfer technology. It is found that the Cu surface coated with the monolayer graphene demonstrate tremendous difference in oxidation pattern and oxidation rate, compared to that coated with the bilayer graphene, which is considered to be originated from the strain-induced linear oxidation channel in monolayer graphene and the intersection of easily-oxidized directions in each layer of bilayer graphene, respectively. We reveal that the defects on the graphene basal plane but not the boundaries are the main oxidation channel for Cu surface under graphene protection. Our finding indi-cates that compared to putting forth efforts to improve the quality of monolayer graphene by reducing defects, depositing multilayer graphene directly on metal is a simple and effective way to enhance the oxidation resistance of graphene-coated metals.     

Representability of actions in the category of (Pre)crossed modules in Leibniz algebras

We investigate the representability of actions in the category of (pre)crossed modules in Leibniz algebras. For this, we construct an actor of a (pre)cat¹-Leibniz algebra and then by using the natural equivalence of the categories of (pre)cat¹-Leibniz algebras and that of (pre)crossed modules, we construct the split extension classifier of the corresponding (pre)crossed module.     

Use of a tree‐structured hierarchical model for estimation of location and uncertainty in multivariate spatial data

Analysis and modeling of spatial data are of considerable interest in many applications. However, the prediction of geographical features from a set of chemical measurements on a set of geographically distinct samples has never been explored. We report a new, tree‐structured hierarchical model for the estimation of geographical location of spatially distributed samples from their chemical measurements. The tree‐structured hierarchical modeling used in this study involves a set of geographic regions stored in a hierarchical tree structure, with each nonterminal node representing a classifier and each terminal node representing a regression model. Once the tree‐structured model is constructed, given a sample with only chemical measurements available, the predicted regional location of the sample is gradually restricted as it is passed through a series of classification steps. The geographic location of the sample can be predicted using a regression model within the terminal subregion. We show that the tree‐structured modeling approach provides reasonable estimates of geographical region and geographic location for surface water samples taken across the entire USA. Further, the location uncertainty, an estimate of a probability that a test sample could be located within a pre‐estimated, joint prediction interval that is much smaller than the terminal subregion, can also be assessed. Copyright © 2014 John Wiley & Sons, Ltd.     

On the type II error in SIMCA method

A novel method for theoretical calculation of the type II (β) error in soft independent modeling by class analogy is proposed. It can be used to compare tentatively predicted and empirically observed results of classification. Such an approach can better characterize model quality and thus improve its validation. Method efficiency is demonstrated on the famous Fisher Iris dataset and on a real‐world example of quality control of packed raw materials in pharmaceutical industry. Copyright © 2014 John Wiley & Sons, Ltd.     

Electronic and structural properties of A Al2Se4 (A=Ag, Cu, Cd, Zn) chalcopyrite semiconductors

We have studied the structural and electronic properties of defect chalcopyrite semiconductors A Al₂Se₄ (A=Ag, Cu, Cd, Zn) using density functional theory (DFT) based first principle technique within tight binding linear muffin-tin orbital (TB-LMTO) method. Our calculated structural parameters such as lattice constants a and c, tetragonal distortion (η=c/2a) are in good agreement with experimental work. Anion displacement parameters, bond lengths and bulk modulus are also calculated. Our band structure calculation suggests that these compounds are direct band gap semiconductors having band gaps 2.40, 2.50, 2.46 and 2.82 eV for A Al₂Se₄ (A=Ag, Cu, Cd, Zn) respectively. Calculated band gaps are in good agreement with other experimental and theoretical works within LDA limitation. We have made a quantitative estimation of the effect of p-d hybridization and structural distortion on the electronic properties. The reduction in band gap due to p-d hybridization is 19.47%, 21.29%, 0% and 0.7% for A Al₂Se₄ (A=Ag, Cu, Cd, Zn) respectively. Increment of the band gap due to structural distortion is 11.62%, 2.45%, 2.92% and 9.30% in case of AgAl₂Se₄, CuAl₂Se₄, CdAl₂Se₄ and ZnAl₂Se₄ respectively. We have also discussed the bond nature of all four compounds.     

Symmetric operators with real defect subspaces of the maximal dimension. Applications to differential operators

Let H be a Hilbert space and let A be a simple symmetric operator in H with equal deficiency indices d:=n_±(A)<∞. We show that if, for all λ in an open interval I⊂R, the dimension of defect subspaces Nλ(A) (=Ker(A^?−λ)) coincides with d, then every self-adjoint extension has no continuous spectrum in I and the point spectrum of is nowhere dense in I. Application of this statement to differential operators makes it possible to generalize the known results by Weidmann to the case of an ordinary differential expression with both singular endpoints and arbitrary equal deficiency indices of the minimal operator.     

高维空间L1范数约束的一类数据稀疏距离测度学习算法与应用

现有一类分类算法通常采用经典欧氏测度描述样本间相似关系,然而欧氏测度不能较好地反映一些数据集样本的内在分布结构,从而影响这些方法对数据的描述能力.提出一种用于改善一类分类器描述性能的高维空间一类数据距离测度学习算法,与已有距离测度学习算法相比,该算法只需提供目标类数据,通过引入样本先验分布正则化项和L1范数惩罚的距离测度稀疏性约束,能有效解决高维空间小样本情况下的一类数据距离测度学习问题,并通过采用分块协调下降算法高效的解决距离测度学习的优化问题.学习的距离测度能容易的嵌入到一类分类器中,仿真实验结果表明采用学习的距离测度能有效改善一类分类器的描述性能,特别能够改善SVDD的描述能力,从而使得一类分类器具有更强的推广能力.     

Investigations on the electron paramagnetic resonance parameters for Mn2+ in the ABF3 fluoroperovskites

The electron paramagnetic resonance (EPR) parameters (g factor, the hyperfine structure constant A and the superhyperfine parameters A' and B') for Mn(2+) in the fluoroperovskites ABF(3) (A=K and Cs; B=Zn, Mg, Cd and Ca) are theoretically investigated from the perturbation formulas of these parameters for a 3d(5) ion under ideal octahedra. In the above treatments, not only the crystal-field mechanism but also the charge transfer mechanism is considered uniformly on the basis of the cluster approach. The theoretical EPR parameters are in good agreement with the experimental data. The charge transfer contribution to the g-shift Δg (≈g-g(s), where g(s)≈2.0023 is the spin-only value) is opposite (positive) in sign and comparable in magnitude to the crystal-field one. Nevertheless, the charge transfer contribution to the hyperfine structure constant shows the same sign and about 10% that of the crystal-field one. So, the conventional argument that the charge transfer contributions to the zero-field splittings are negligible for 3d(5) ions under low symmetrically distorted fluorine octahedra is proved no longer valid for the Δg analysis of ABF(3):Mn(2+) in view of the dominant second-order charge transfer perturbation terms. The unpaired spin densities of the fluorine 2s, 2p σ and 2p π orbitals are determined from the quantitative dependences upon the related molecular orbital coefficients, rather than obtained by fitting the observed superhyperfine parameters in the previous works.     

Bauschinger and size effects in thin-film plasticity due to defect-energy of geometrical necessary dislocations

The Bauschinger and size effects in the thinfilm plasticity theory arising from the defect-energy of geometrically necessary dislocations (GNDs) are analytically investigated in this paper. Firstly, this defect-energy is deduced based on the elastic interactions of coupling dislocations (or pile-ups) moving on the closed neighboring slip plane. This energy is a quadratic function of the GNDs density, and includes an elastic interaction coefficient and an energetic length scale L. By incorporating it into the work- conjugate strain gradient plasticity theory of Gurtin, an energetic stress associated with this defect energy is obtained, which just plays the role of back stress in the kinematic hardening model. Then this back-stress hardening model is used to investigate the Bauschinger and size effects in the tension problem of single crystal Al films with passivation layers. The tension stress in the film shows a reverse dependence on the film thickness h. By comparing it with discrete-dislocation simulation results, the length scale L is determined, which is just several slip plane spacing, and accords well with our physical interpretation for the defect- energy. The Bauschinger effect after unloading is analyzed by combining this back-stress hardening model with a friction model. The effects of film thickness and pre-strain on the reversed plastic strain after unloading are quantified and qualitatively compared with experiment results.