结合深度学习与支持向量机的金属零件识别

目的 在视觉引导的工业机器人自动拾取研究中，关键技术难点之一是机器人抓取目标区域的识别问题。特别是金属零件，其表面的反光、随意摆放时相互遮挡等非结构化因素都给抓取区域的识别带来巨大的挑战。因此，本文提出一种结合深度学习和支持向量机的抓取区域识别方法。方法 分别提取抓取区域的方向梯度直方图（HOG）和局部二进制模式（LBP）特征，利用主成分分析法（PCA）对融合后的特征进行降维，以此来训练支持向量机（SVM）分类器。通过训练Mask R-CNN（regions with convolutional neural network）神经网络完成抓取区域的初步分割。然后利用SVM对Mask R-CNN识别的抓取区域进行二次分类，完成对干扰区域的剔除。最后计算掩码完成实例分割，以此达到对抓取区域的精确识别。结果 对于随机摆放的铜质金属零件，本文算法与单一的Mask R-CNN及多特征融合的SVM算法就识别准确率、错检率、漏检率3个指标进行了比较，结果表明本文算法在识别准确率上较Mask R-CNN和SVM算法分别提高了7%和25%，同时有效降低了错检率与漏检率。结论 本文算法结合了Mask R-CNN与SVM两种方法，对于反光和遮挡情况具有一定的鲁棒性，同时有效地提升了目标识别的准确率。     

Study on the long-term behaviour of glass fibre in the tensile stress field

Glass fibre is a random network structure composed of [SiO₄] tetrahedra. The structure contains a large number of defects, which act as crack initiation points. Under tensile stress, cracks undergo crack initiation, stable propagation, failure propagation, and fracture, and the stress that begins after unstable propagation is called the critical fracture stress. When the stress is less than the critical value, the crack is subject to the force of chemical bonds during the crack propagation process, and crack arrest occurs. When the stress is greater than the critical value, the glass fibre will undergo destructive fracture. In this paper, long-term tensile tests were carried out on glass fibre and a glass fibre composite under different constant tensile stress conditions. The fracture times of the glass fibre and glass fibre composite under different tensile stresses were obtained, the critical fracture stress of glass fibre was inferred, and the fracture mechanism was explained.     

Microwave irradiated Carrageenan-Guar gum micro-porous IPN: a novel material for isotropic tissue scaffolding

In this study, a novel micro porous honeycomb structured Crg-GG-IPN material was incepted to be applicable as scaffold and accomplished. The hydrophilicity was confirmed by FT-IR and OCA. Amorphous nature and micro-rough surface were confirmed by XRD and AFM. Void fraction was 0.61?±?0.04. Void space, hemocompatibility and platelet adhesion were captured by SEM. Degradability of the material was confirmed by in-vitro degradation study. Incision method using mice model was a clear evidence for cell attachment and non-toxicity and was confirmed from hematology and histopathology. Thus, it appears that Crg-GG scaffolds can be useful as wound healing material for clinical applications.     

GreyFan:一种Wi-Fi隐蔽信道攻击框架

针对现实网络中诸如侧信道攻击、HID 攻击等传统的物理攻击，物理隔离被认为是一种较为彻底的抵御网络攻击的安全防护手段。2018 年，业界首次提出了一种物理隔离环境下的 Wi-Fi 隐蔽信道方法——Ghost Tunnel，即在Wi-Fi尚未连接的状态下，无线AP可成功将数据传给发起连接请求的计算机。提出了一种基于Ghost Tunnel方法的攻击框架—— GreyFan，利用该攻击框架攻击者可以对未连接Wi-Fi的用户实施无感知攻击，如文件隐蔽传输、任意代码执行等，并分析了相应的防御技术。     

Exploiting Bayesian networks for fault isolation: A diagnostic case study of diesel fuel injection system

Fault isolation is known to be a challenging problem in machinery troubleshooting. It is not only because the isolation of multiple faults contains considerable number of uncertainties due to the strong correlation and coupling between different faults, but often massive prior knowledge is needed as well. This paper presents a Bayesian network-based approach for fault isolation in the presence of the uncertainties. Various faults and symptoms are parameterized using state variables, or the so-called nodes in Bayesian networks (BNs). Probabilistically causality between a fault and a symptom and its quantization are described respectively by a directed edge and conditional probability. To reduce the qualitative and quantitative knowledge needed, particular considerations are given to the simplification of Bayesian networks structures and conditional probability expressions using rough sets and noisy-OR/MAX model, respectively. By adopting the simplified approach, symptoms under multiple-fault are decoupled into the ones under every single fault, while the quantity of the conditional probabilities is simplified into the linear form of the faults quantity. Prior knowledge needed in Bayesian network-based diagnostic model is reduced significantly, which decreases the complexity in establishing and applying this diagnosis model. The computational efficiency is improved accordingly in the simplified BN model, after eliminating the redundant symptoms. The fault isolation methodology is illustrated through an example of diesel engine fuel injection system to verify the developed model.     

网络匿名扫描系统设计及优化

针对网络扫描工具在进行扫描时面临的溯源问题，提出了一种匿名网络扫描系统。首先将匿名系统与网络扫描工具结合以实现匿名扫描；然后在现有匿名系统的基础上实现了该系统的本地私有化；接着通过流量分析发现，Nmap的多进程扫描因为代理链的原因会变成单进程扫描而导致其扫描扫描性能较低；最后提出了一种基于多Namp进程并发的性能优化方案，将总体扫描任务分割为多个扫描任务，并分配给多个单独的Nmap进程并行运行。实验结果表明，该性能优化方案的扫描时延接近正常扫描情况下的时延，达到了提高匿名扫描系统性能的目的。因此，该优化后的网络匿名扫描系统在阻碍溯源的同时提升了扫描效率。     

Adaptive neural network command filtered backstepping control of pure‐feedback systems in presence of full state constraints

An adaptive neural network (NN) command filtered backstepping control is proposed for the pure‐feedback system subjected to time‐varying output/stated constraints. By introducing a one‐to‐one nonlinear mapping, the obstacle caused by full stated constraints is conquered. The adaptive control law is constructed by command filtered backstepping technology and radial basis function NNs, where only one learning parameter needs to be updated online. The stability analysis via nonlinear small‐gain theorem shows that all the signals in closed‐loop system are semiglobal uniformly ultimately bounded. The simulation examples demonstrate the effectiveness of the proposed control scheme.     

基于深度学习的彩色以及近红外图像去马赛克

单传感器捕获的彩色-近红外（RGB-NIR）图像存在光谱干扰，从而导致重建出的标准彩色图像（RGB）图像与近红外（NIR）图像存在色彩失真以及细节信息模糊。针对这个问题提出一种基于深度学习的去马赛克方法，通过引入跳远连接与稠密连接解决了梯度消失和梯度弥散问题，使得网络更容易训练，并且提升了网络的拟合能力。首先，用浅层特征提取层提取了马赛克图像的像素相关性以及通道相关性等低级特征；然后，将得到的浅层特征图输入到连续多个的残差稠密块以提取专门针对去马赛克的高级语义特征；其次，为充分利用低级特征与高级特征，将多个残差稠密块提取到的特征进行组合；最后，通过全局跳远连接恢复最终的RGB-NIR图像。在深度学习框架Tensorflow上使用公共的图像与视觉表示组（IVRG）数据集、有植被的户外多光谱图像（OMSIV）数据集和森林（Forest）三个公开数据集进行实验。实验结果表明，所提方法优于基于多级自适应残差插值、基于卷积卷积和神经神经网络以及基于深度残差U型网络的主流的RGB-NIR图像去马赛克方法。     

基于k-shell的复杂网络最短路径近似算法

复杂网络最短路径经典算法的处理效率较低，不适用于大规模复杂网络，而现有近似算法通用性有限，且计算准确率不理想，不能满足规模日益扩大的复杂网络中的最短路径计算需求。针对于此，提出基于[k]-shell的复杂网络最短路径近似算法。算法利用节点的[k]-shell值进行网络划分并引导搜索路径，利用超点聚合处理[k]-shell子网来降低路径搜索中节点和连边的规模，通过在路径搜索过程使用双向搜索树方法提高算法的计算效率和准确率。实验结果表明，算法通用性较好，在现实与仿真大规模复杂网络中均具有较高的计算效率和准确率。