大数据系统和分析技术综述

首先根据处理形式的不同,介绍了不同形式数据的特征和各自的典型应用场景以及相应的代表性处理系统,总结了大数据处理系统的三大发展趋势;随后,对系统支撑下的大数据分析技术和应用(包括深度学习、知识计算、社会计算与可视化等)进行了简要综述,总结了各种技术在大数据分析理解过程中的关键作用;最后梳理了大数据处理和分析面临的数据复杂性、计算复杂性和系统复杂性挑战,并逐一提出了可能的应对之策.     

基于多渐消因子强跟踪UKF 和约束AR模型的故障估计与预测

针对非线性系统中不可观测故障参数估计和预测问题, 提出一种基于多重渐消因子强跟踪无迹卡尔曼滤波(MSTUKF) 的状态和参数联合估计法, 通过引入多重渐消因子增强了对变化函数未知的故障参数的跟踪能力. 对于得到的故障参数估计值, 利用递推最小二乘法更新约束AR预测模型, 从而实现故障参数的在线估计与预测. 仿真结果表明, MSTUKF方法在故障参数估计精度上优于UKF 和单渐消因子强跟踪UKF, 约束AR模型的预测精度高于无约束条件下的预测精度.

     

Applications of laser Raman microprobe spectroscopy to the characterization of coals and cokes

Optical microscopy is widely used in the characterization of coals and cokes. This Paper shows that the laser Raman microprobe (MOLE) which combines an optical microscope and a Raman spectrometer can provide useful additional information. Three main areas were investigated: identification of minerals in coal and coke; structural characterization of coals and cokes; and the interaction of inorganic additives and coal. Where possible, the results were compared with conventional optical microscopy measurements whereby it was shown that the optical texture and Raman spectra of cokes are not closely related. The Raman spectra of high temperature cokes were used to estimate the size of microcrystalline regions.     

金陵电厂proFiBuS总线控制系统维护经验总结

华能金陵超超临界电厂是中国第一家在主辅控区域同时实行了PROFIBUS现场总线控制的百万机组级、全数字化电厂,现场大量使用了与传统DCS控制不同的控制网络、现场智能化设备,维护手段和方式也有别于传统的控制系统。这为数字智能化电厂的建设打下了基础,具有重要的示范意义。本文以华能金陵电厂脱硫系统的总线控制系统为例,介绍了总线方面的相关概念和知识,通过对总线功能作用的阐述,以现场总线智能设备为切入点,结合平时的维护检修工作,对PROFIBUS的优缺点进行了分析总结,并提出了今后维护工作的注意事项。     

基于字的分布表征的汉语基本块识别

汉语的基本块识别是汉语句法语义自动分析中的重要任务之一。传统的方法大多数直接将汉语基本块识别任务转化成词层面的一个序列标注问题,采用CRF模型来处理。虽然,在许多评测中得到最好的结果,但基于词为标注单位,在实用中受限于自动分词系统以及汉语词特征的稀疏性。为此,该文给出了一种以字为标注单位,以字为原始输入层,来构建汉语的基本块识别的深层神经网络模型,并通过无监督方法,学习到字的C&W和word2vec两种分布表征,将其作为深层神经网络模型的字的表示层的初始输入参数来强化模型参数的训练。实验结果表明,使用五层神经网络模型,以[-3,3]窗口的字的word2vec分布表征,其准确率、召回率和F值分别达到80.74%,73.80%和77.12%,这比基于字的CRF高出约5%。这表明深层神经网络模型在汉语的基本块识别中是有作用的。     

一种可扩展的动态数据持有性证明方案

提出了一种能保护数据隐私的动态数据持有性证明方案, 并对方案的安全性进行了证明, 性能分析表明该方案具有低存储、计算和传输负载的特点。通过扩展实现了对公开可验证和多副本检查的支持, 扩展方案的特点在于公开验证中不需要可信第三方的参与, 而多副本检查可以方便地实现对所有副本的批量检查。     

Proton diffusion in γ-manganese dioxide

Deconvolution of the electrolytic manganese dioxide (EMD) discharge curve has indicated the presence of a number of energetically different reduction processes. This has been used to determine the contribution of each reduction process to the total discharge. Using step potential electrochemical spectroscopy (SPECS), the i-t data were modelled as the sum of the discharge of the individual reduction processes. From this, A√ D for each reduction process as a function of degree of discharge was determined. The maximum A√ D values for each process ranged from 2.3×10⁻² to 4.0×10⁻⁴ cm³ s^−1/2 g⁻¹ values are consistent with previously reported values for A√ D, although in this case we have determined values for the entire compositional range.     

Fabrication and optical characterization of poly(2,5‐di‐n‐butoxyphenylene) nanofibril arrays

Anodic aluminum oxide (AAO) membrane can be used as template for the synthesized nanostructures. In this article, we have prepared the AAO membrane by using electrooxidation of aluminum substrate in phosphoric acid, and fabricated poly(2,5‐di‐n‐butoxyphenylene) (BuO–PPP) nanofibril arrays by oxidative coupling polymerization of 1,4‐di‐n‐butoxybenzene (DBB) within the pores of the AAO template membrane. The detailed molecular structure of the polymer nanofibrils was characterized by using infrared and ¹H nuclear magnetic resonance spectra, and estimated to consist of almost equal fractions of 1,4‐ and 1,3‐ linkages. We have used transmission electron microscopy, scanning electron microscopy, and atom force microscopy to confirm the morphologies and images of the AAO template membrane and the fabricated nanometer scale of BuO–PPP nanofibril arrays. The experimental results demonstrated that the pores of the AAO membrane were regular and uniform, and parallel each other, and the BuO–PPP chains in the narrowest template‐synthesized nanofibrils were oriented parallel to the porous axes of the AAO membrane and perpendicular to the surface of the aluminum substrate. The polymer chain orientation was partially responsible for the enhanced conductivity. The ultraviolet absorption spectrum of the BuO–PPP nanofibril arrays shown that the polymer contains a better extended π‐conjugation system along poly‐(p‐phenylene) backbone, which resulted in longer wavelength shift of the absorption band, the absorption maxima were located at 258 nm (E₁ absorption band) and 332 nm (E₂ absorption band), respectively. Photoluminescence spectrum of the BuO–PPP nanofibril arrays exhibited a blue emission. © 2003 Wiley Periodicals, Inc. J Appl Polym Sci 91: 425–430, 2004     

基于EPR态的量子代理签名方案*

提出了一种量子代理签名方案,利用量子力学中Einstein-Podolsky-Rosen(EPR)的纠缠特性并结合经典编码方法来实现对量子比特串的签名和验证。在本方案中,原始签名人可以将自己的签名权委托给代理签名人,而量子密钥分配和一次一密保证了新方案的无条件安全性。研究表明新方案满足不可伪造性、不可否认性和可追踪性。     

模块化免疫神经网络平衡态检测模型*

基于神经网络原理、免疫系统和遗传算法的相关机理,构造了一个网络安全平衡器。该平衡器建立了抗原与抗体平衡态检测的数学模型、抗原与抗体的促进和抑制函数,提出了安全平衡态的概念,给出了抗原与抗体浓度的计算公式和模块化多层分类处理模型,利用促进、抑制函数和遗传算法适应度函数使已受攻击的网络再次达到一种新的安全平衡状态,这为网络安全提供了一个新的途径。理论证明网络安全方案是有效的。