面向远程诊断的AI-ESTATE智能诊断系统研究动态

通过封装服务、规范标准服务,AI-EATATE为诊断系统的客户端和应用对象的扩展、测试与诊断的分离提供了条件,使其更适合远程控制的诊断系统的构建。论文分析了武器装备远程故障诊断对知识重用与共享、软件互操作的需求,提出了面向远程诊断的AI-ESTTAE智能诊断系统构建的方法;研究了AI-ESTATE智能诊断系统的基本结构和框架,讨论了AI-ESTATE在发展过程中其服务的变化对开放、远程智能诊断系统构建的影响;最新的AI-ESTATE标准规范的服务更加强了对诊断过程的控制和效果评估,论文采用状态图的方式分析了最新的AI-ESTTAE服务在远程故障诊断过程的作用,为符合AI-ESTATE标准的远程故障诊断系统的开发提供了方法。     

面向服务的AI-ESTATE故障诊断系统建模

测试和诊断的融合可以提高武器系统的维护效率,但是随着武器装备的升级改造就带来了原有的诊断知识难共享、诊断推理软件难以互操作、难重用,诊断功能难扩展的问题;AI-ESTATE标准规范了诊断知识和数据的标准化描述和诊断推理机的服务接口,为知识互换和软件可移植提供了条件;论文研究了面向服务的AI-ESTATE开放式故障诊断系统的体系结构和信息传递模式,分析了推理机模型管理服务和互操作服务;然后运用静态诊断模型服务关系图和UML时序图分析了AI-ESTATE服务在诊断模型编辑、索引等功能实现中的交换流程,运用动态诊断模型的UML时序图分析了AI-ESTATE互操作服务在故障诊断系统运行中推理机与其它组件的动态交互过程,为开放式AI-ESTATE智能诊断系统的开发奠定了基础。     

不确定海洋环境中基于主成分量的稳健检测器

传统的信号检测算法在不确定的海洋环境中性能出现下降。基于贝叶斯原理的最优检测算法可以实现对不确定海洋环境中信号的有效检测,但是其突出问题是计算量较大。本文提出了一种基于主成分量分析的稳健信号检测器,该检测器利用贝叶斯原理将环境先验信息引入到检测算法中,同时使用主成分量分析方法来降低运算量,实现了对信号的快速有效检测。分别使用标准失配海洋模型和海上实测数据进行了计算机仿真和实验验证,结果表明:(1)基于主成分量的稳健信号检测器检测性能达到最优贝叶斯检测器的效果。(2)本文方法在线运算速度是贝叶斯最优检测器的5^一8倍。(3)环境先验信息失配的情况下,扩大海洋环境参数模型的不确定度范围有助于提高检测性能。      

基于MapReduce的贝叶斯网络参数学习方法

期望最大化(Expectation Maximization,EM)算法常被应用于贝叶斯网络参数学习过程,但在处理海量数据时由于迭代计算过程的复杂性和处理器、内存等资源的限制,该算法的效能受到极大影响;通过对大数据环境下传统线性贝叶斯网络参数学习方法计算复杂性瓶颈问题的研究,提出了基于MapReduce平台的贝叶斯网络并行期望最大化(Parallel Expectation Maximization,PEM)参数学习算法;利用不完备训练样本集,对态势评估贝叶斯网络进行参数学习;仿真结果表明:在大数据条件下PEM算法能够准确的学习网络参数,同时有效减少参数学习所需时间且具有较好的可拓展性。     

基于移动机器人和单目视觉的姿态测量方法

在单目视觉的姿态测量工作中,传统卷积神经网络在模糊场景及复杂场景下存在准确度大幅降低的问题,为此提出了一种基于胶囊网络与贝叶斯网络相结合的深度学习模型,在此基础上提出了基于移动机器人与单目视觉的姿态测量方法。采用新型胶囊网络对单目视觉目标的重要关节点进行空间定位;设计了简单的贝叶斯网络学习算法,通过贝叶斯网络推理出关节点的空间姿态;在复杂的人体姿态测量数据集上完成了验证实验,结果表明实现了较好的测量准确度,在复杂场景下依然保持了较好的准确度,在室内与室外环境下的F1-measure值分别达到0.9和0.78。     

基于Bayesian Network列控系统超速故障的分析

列控系统非常复杂,在对其进行故障分析时,自身的特质决定了用传统的故障树在对其进行故障分析时会产生局限性;为此文中引入了贝叶斯网络技术,充分利用其推理算法成熟、理论基础非常完备、学习能力非常强的优势,将事件树中各环节的故障树用贝叶斯网络进行描述;利用贝叶斯网络工具箱(BNT)对列车超速故障进行因果和诊断推理分析,通过数值计算结果得出所在不同场景下故障发生的主要原因,并提出减少故障发生概率相应的措施。     

社交网络中基于贝叶斯和半环代数模型的节点影响力计算机理

本文综合考虑网络结构及节点间的互动等关键因素, 提出了一种节点影响力分布式计算机理. 首先根据节点交互行为在时域上的自相似特性, 运用带折扣因子的贝叶斯模型计算节点间的直接影响力; 然后运用半环模型来分析节点间接影响力的聚合; 最后根据社交网络的小世界性质及传播门限, 综上计算出节点的综合影响力. 仿真结果表明, 本文给出的模型能有效抑制虚假粉丝导致的节点影响力波动, 消除了虚假粉丝的出现对节点影响力计算带来的干扰, 从中选择影响力高的若干节点作为传播源节点, 可以将信息传播到更多数目的节点, 促进了信息在社交网络中的传播. 关键词： 社交网络 影响力 贝叶斯 半环代数     

基于位移特征与随机森林的表情识别方法研究

为了有效地对图像序列进行面部表情识别,提出一种基于主动形状模型(active shape model,ASM)结合Lucas-Kanade(LK)光流法的方法提取位移特征,采用随机森林分类器对提取到的位移特征进行分类与识别。在Extended Cohn-Kanade(CK+)人脸表情数据库上的实验表明,该特征提取方法能够很好地描述图像序列中所包含的表情信息和特征点运动变化信息,比常用的K-近邻、贝叶斯网络和支持向量机等分类器所表现出来的效果要好,其识别率达到95.1%。     

基于SiPLS特征提取和信息融合的汽油中乙醇含量的多光谱检测

采用紫外可见(ultraviolet/visible, UV-Vis)光谱技术和近红外(near-infrared, NIR)光谱技术及信息融合技术对乙醇汽油中乙醇含量进行了检测。首先采用组合区间偏最小二乘(synergy interval PLS, SiPLS)算法作为特征提取方法,分别建立了基于UV-Vis和NIR光谱的偏最小二乘(PLS)回归模型;再根据油品的实际情况,运用信息融合理论将UV-Vis和NIR光谱信息进行融合,建立了数据级融合(low level data fusion, LLDF)和特征级融合(mid-level data fusion, MLDF)模型,并与单谱源模型效果进行了比较,确定了最优模型为数据级融合后再进行矢量归一化的模型(LLDF-VN1);最后分别用高乙醇含量样品和市售汽油样品的光谱数据对该最优模型进行了通用性检验。结果表明：UV-Vis和NIR光谱数据单独建模均能很好的检测并提供较好的预测结果;而UV-Vis和NIR光谱数据直接融合在基于校正集的回归模型中效果最好,其校正集相关系数r_c=0.999 9,校正集交叉验证均方差RMSECV=0.125 8,校正集整体评价偏差Bias_c=0.000 6;而采用数据级融合后再进行矢量归一化的模型(LLDF-VN1)的预测效果为最佳,其r_p=0.999 1, RMSEP=0.352 7, Bias_p=-0.073 8;自配溶液对最优模型(LLDF-VN1)的通用性验证中,r_p=0.999 7, RMSEP=0.329 1, Bias_p=0.102 2;市售汽油对最优模型(LLDF-VN1)的通用性验证中,r_p=0.990 1, RMSEP=0.892 7, Bias_p=0.675 1。实验结果说明通过将UV-Vis和NIR光谱信息进行数据级融合可以快速、准确的检测出乙醇汽油中乙醇的含量,并能实现乙醇浓度的宽范围检测,为进一步实现混合油品中物质的快速检测奠定了基础。     

近、中红外光谱法融合判定葡萄酒产地

葡萄酒带有浓厚的葡萄原产地地域特点与个性,快速准确地判别葡萄酒原产地具有重要意义,感官评定的方法存在一定的局限性。提出用贝叶斯信息融合技术将葡萄酒样品的近红外透射光谱及中红外衰减全反射光谱联立进行葡萄酒原产地判别的方法。分别用近、中红外光谱仪采集来自中国四个不同葡萄主栽产地(河北怀来、山东烟台、甘肃、河北昌黎)的153个葡萄酒样品的近红外透射光谱和中红外衰减全反射光谱,然后用偏最小二乘判别分析法(PLS-DA)分别建立基于近红外光谱和中红外光谱的葡萄酒产区判别模型;该模型输出的节点值归一化后作为所有样品分属每一类别的先验概率,代入Bayes判别公式得到后验概率,根据此概率判断样品的新类别属性,即用贝叶斯信息融合技术实现了两种判别结果的修正决策。近红外和中红外融合后的模型结果为：十次随机划分建模集和检验集,四产区葡萄酒判别模型建模集的平均准确率由78.21%(近红外)和82.57%(中红外)变为融合后的87.11%,检验集平均准确率由82.50%(近红外)和81.98%(中红外)变为融合后的90.87%,均优于单独采用一种光谱技术的判别结果。实验结果表明：信息融合技术有助于模型判别效果的提高,采用近、中红外光谱的贝叶斯信息融合技术对葡萄酒原产地进行快速识别是可行的。     

Are there anomalous Z fermion couplings?

The couplings of the fermions to the Z boson are of great importance in establishing the validity of the Standard Model and in looking for physics beyond it. The couplings of the b-quark to the Z boson have been the subject of much experimental study and theoretical interpretation. The apparent excess in the value of , the ratio of the partial width of the Z boson to to its total hadronic width, above the Standard Model expectation reported a few years ago has now become much less significant. However, the measurements of the pole forward-backward asymmetry for b-quarks at the Z pole and of the polarisation parameter , obtained using a polarised electron beam, have improved considerably in accuracy. The latest data are examined and values of the vector and axial-vector b-quark and c-quark couplings to the Z are extracted. The left and right handed couplings are also extracted. It is found that whereas the c-quark couplings are compatible with the Standard Model, those of the b-quark data are only compatible with the Standard Model at about the 1% level. In addition, the individual lepton couplings are extracted and the degree to which the data support the hypothesis of lepton universality is discussed. The sensitivity of the limits from electroweak fits to the Higgs boson mass to these data is examined. Received: 21 December 1998 / Revised version: 12 February 1999 / Published online: 15 April 1999     

The probable fate of the Standard Model

Extrapolating the Standard Model to high scales using the renormalisation group, three possibilities arise, depending on the mass of the Higgs boson: if the Higgs mass is large enough the Higgs self-coupling may blow up, entailing some new non-perturbative dynamics; if the Higgs mass is small the effective potential of the Standard Model may reveal an instability; or the Standard Model may survive all the way to the Planck scale for an intermediate range of Higgs masses. This latter case does not necessarily require stability at all times, but includes the possibility of a metastable vacuum which has not yet decayed. We evaluate the relative likelihoods of these possibilities, on the basis of a global fit to the Standard Model made using the Gfitter package. This uses the information about the Higgs mass available directly from Higgs searches at LEP and now the Tevatron, and indirectly from precision electroweak data. We find that the ‘blow-up’ scenario is disfavoured at the 99% confidence level (96% without the Tevatron exclusion), whereas the ‘survival’ and possible ‘metastable’ scenarios remain plausible. A future measurement of the mass of the Higgs boson could reveal the fate of the Standard Model.     

Flavor changing neutral currents in the Lee–Wick Standard Model

Recently an extension of the Standard Model (the Lee–Wick Standard Model) based on ideas of Lee and Wick (LW) was introduced. It does not contain quadratic divergences in the Higgs mass and hence solves the hierarchy puzzle. The LW Standard Model contains new heavy LW-resonances at the TeV scale that decay to ordinary particles. In this Letter we examine in more detail the flavor structure of the theory. We integrate out the heavy LW-fermions at tree level and find that this induces flavor changing Z-boson couplings. However, these flavor changing neutral currents are acceptably small since they are automatically suppressed by small Yukawa couplings. This is the case even though the theory does not satisfy the principle of minimal flavor violation. New couplings of the charged W-bosons to quarks and leptons are also induced. We also integrate out the LW–Higgs and examine the four-fermion operators induced.     

Final results from DELPHI on the searches for SM and MSSM neutral Higgs bosons

These final results from DELPHI searches for the Standard Model (SM) Higgs boson, together with benchmark scans of the Minimal Supersymmetric Standard Model (MSSM) neutral Higgs bosons, used data taken at centre-of-mass energies between 200 and 209 GeV with a total integrated luminosity of 224 pb^-1. The data from 192 to 202 GeV are reanalysed with improved b-tagging for MSSM final states decaying to four b-quarks. The 95% confidence level lower mass bound on the Standard Model Higgs boson is 114.1 GeV/c ². Limits are also given on the lightest scalar and pseudo-scalar Higgs bosons of the MSSM.Received: 7 March 2003, Revised: 30 September 2003, Published online: 3 December 2003     

Tests of the standard model and constraints on new physics from measurements of fermion-pair production at 130–172 GeV at LEP

Production of events with hadronic and leptonic final states has been measured in collisions at centre-of-mass energies of 130–172 GeV, using the OPAL detector at LEP. Cross-sections and leptonic forward-backward asymmetries are presented, both including and excluding the dominant production of radiative Z events, and compared to Standard Model expectations. The ratio of the cross-section for production to the hadronic cross-section has been measured. In a model-independent fit to the Z lineshape, the data have been used to obtain an improved precision on the measurement of -Z interference. The energy dependence of has been investigated. The measurements have also been used to obtain limits on extensions of the Standard Model described by effective four-fermion contact interactions, to search for -channel contributions from new massive particles and to place limits on gaugino pair production with subsequent decay of the gaugino into a light gluino and a quark pair. Received: 30 July 1997 / Published online: 10 March 1998     

模型飞行验证系统测控与信息传输技术研究

模型飞行验证系统是为更加有效地开展模型自由飞试验研究而设计的通用试验系统,测控(TT C)与信息传输系统是其信息保障的核心。本文在充分考虑模型自由飞试验特点的基础上,重点研究了模型飞行验证系统研制过程中的测控与信息传输技术,介绍了其工作原理及实现方法,并对部分结果进行分析。     

\mathrm{W^+W^-} production and triple gauge boson couplings at LEP energies up to 183 GeV

A study of W-pair production in annihilations at LEP is presented, based on candidates corresponding to an integrated luminosity of 57 pb at GeV. Assuming that the angular distributions of the W-pair production and decay, as well as their branching fractions, are described by the Standard Model, the W-pair production cross-section is measured to be pb. Assuming lepton universality and combining with our results from lower centre-of-mass energies, the W branching fraction to hadrons is determined to be %. The number of W-pair candidates and the angular distributions for each final state (, are used to determine the triple gauge boson couplings. After combining these values with our results from lower centre-of-mass energies we obtain , and , where the errors include both statistical and systematic uncertainties and each coupling is determined by setting the other two couplings to the Standard Model value. The fraction of W bosons produced with a longitudinal polarisation is measured to be . All these measurements are consistent with the Standard Model expectations. Received: 23 October 1998 / Published online: 1 March 1999     

Highlights on searches for supersymmetry and exotic models

In this review, we present highlight results of the first three years of the LHC running on searches for new physics beyond the Standard Model. The excellent performance of the LHC machine and detectors has provided a large, high-quality dataset, mainly proton–proton interactions at a centre of mass energy of 7 TeV (collected in 2010 and 2011) and 8 TeV (collected in 2012). This allowed the experiments to test the Standard Model at the highest available energy and to search for new phenomena in a considerably enlarged phase space compared to previous colliders.