移动P2P网络双向匿名通信机制设计

无线链路不稳定和恶意节点不合作等因素会影响移动P2P网络匿名机制的实现。为解决该问题,通过改进移动P2P网络结构,提出一种可以随网络状况自适应变化的NCS-Crowds匿名算法,综合运用假名映射和掩饰流等匿名技术,设计系统架构演进架构下的双向匿名通信机制。理论分析和仿真结果证明,该机制能够根据网络状况在转发成功率和开销代价之间取得较好的折中,实现双向匿名通信。     

燃料电池汽车TTCAN网络的调度设计

针对目前燃料电池汽车应用层协议不统一的情况,该文以自主研发的燃料电池中巴车为对象,基于SAEJ1939协议,利用均匀装载(AL)调度算法实现了其动力总成系统TTCAN协议的调度设计,该协议对其各节点源地址分配、优先级和参数组定义做出明确规定。通过可调度性、实时性、周期抖动、总线负载率和发送延迟的分析和测试,表明该协议是高效可行的,能满足燃料电池汽车实时控制的要求。     

Rolling Contact Fatigue Properties of SAE 8620 Steel after Case Carburizing

Rolling contact fatigue(RCF)properties of SAE 8620 steel after case carburizing have been investigated under two contact stresses of 4.0and 5.5GPa.Results show that the RCF life ranges from 2.5×10~6 to 3×10~7 cycles under the contact stress of 5.5GPa,while it can be more than 1×10~8 cycles under the contact stress of 4.0GPa.The rated fatigue life L_(10)(lives with the 10%failure)is also drastically shortened from 9.8×10~6 to 5.4×10~5 cycles when the contact stress is increased from 4.0to 5.5GPa.Theoretical calculations and fractographs show that the maximum shear stress and the contact area increase with increasing the contact stress,making RCF tend to occur earlier.     

3GPP LTE系统安全性研究

长期演进（LTE）系统代表TUMTS标准的最新发展方向,作为一种纯分组系统,它具有全新的无线接口和体系结构,其频谱效率是HSDPA／HSUPA的2倍。本文分析了LTE系统安全的主要内容,提出TLTE用户安全的体系结构,描述TIP网络的NDS体系结构,介绍了隧道模式下的ESP保护原理。     

Computational Thermodynamics Application on the Calcium Inclusion Treatment of SAE 8620 Steel

A thermodynamic study was carried out to evaluate the contents of Al, Ca, S and O at which liquid or partially liquid inclusions are formed to optimize the castability of SAE 8620 steel. Samples of steel secondary treatment (Ladle) and continuous‐casting billets (Tundish) were obtained from an electric steel mill. Entry data for this study were obtained in heats carried out in the steel plant. Furthermore, analyses were performed with a Scanning Electron Microscope (SEM) and an Energy‐Dispersive Spectrometer (EDS), as well as chemical analysis of both steel and total oxygen. In the Ladle sample, after calcium addition, inclusions had concentrations between 30 and 45% CaO and consisted mainly of calcium‐aluminates with MgO content of 10%. In the Tundish sample, inclusions contained a high percentage of alumina. Combining the characterization of inclusions with a computational thermodynamics model, the obtained results showed that the ideal range of calcium may vary between 10 and 14 ppm to form liquid inclusions with a minimum CaS content for the O, S and Al levels from SAE 8620 in this process stage. An analysis of Mg influence on the formation of calcium‐aluminates was also carried out. For contents of Mg found in aluminium‐killed steels treated with Ca, Mg does not significantly influence the formation of calcium‐aluminates inclusions. Higher calcium contents can be taken into account, depending on both the process type in steel production and the demanded product properties. The present study shows that the combined application of both inclusions characterization techniques and computational thermodynamics can drastically reduce the empiricism regarding the development of processes for the control of non‐metallic inclusions in steels. Thus, this research can contribute to increase the efficiency of manufacturing processes, improve the quality of existent products and make possible the economic development of new ones.     

The lubricated wear of ceramics : Part II: The wear and friction of silicon nitride and steel in the presence of mineral oil or an ester based lubricant

The friction and wear characteristics of combinations of silicon nitride, alumina and AISI 52100 steel in the presence of mineral oil containing anti-wear, dispersant and detergent additives have been investigated in a tri-pin-on-disc machine. The tests were carried out at a nominal temperature of 100°C for a range of sliding speeds, loads and total sliding distances. In Part II of this two-part paper a comparison will be made between the tribological performance of these sliding pairs of materials in mineral oil and ester based lubricant environments. The results of the investigation showed that the alumina performed relatively poorly under these test conditions, whereas silicon nitride showed good potential as an improved wear resisting material compared with 52100 steel. Wear factors of the order of 10⁻¹⁰ mm³/Nm were deduced for the alumina, while values as low as 10⁻¹¹ mm³/Nm were typical of the silicon nitride sliding against 52100 steel discs. The alumina pins wore by a process of brittle fracture at the surface, whereas the silicon nitride pins wore primarily by a tribochemical polishing mechanism. The rate of tribo-chemical wear was found to be proportional to the nominal contact area.     

FATIGUE CRACK GROWTH UNDER MIXED-MODE I AND II LOADING

Abstract— Mixed-mode fatigue crack growth has been studied using four point bend specimens under asymmetric loads. A detailed finite element analysis provides the stress intensity factors for curved cracks under different mixed-mode load conditions. Both fatigue crack growth direction and crack growth rate are studied. The maximum tangential stress and the minimum strain energy density criteria were found to provide satisfactory predictions of the crack growth directions. An effective stress intensity factor was used to correlate the fatigue crack growth rates successfully. It is found that the use of mode I fatigue crack growth rate properties results in a conservative crack growth rate prediction for mixed-mode load conditions.     

基于SAE和GNDO-SVM的脑电信号情绪识别

情感计算是现代人机交互中的关键问题, 随着人工智能的发展, 基于脑电信号(electroencephalogram, EEG)的情绪识别已经成为重要的研究方向. 为了提高情绪识别的分类精度, 本研究引入堆叠自动编码器(stacked auto-encoder, SAE)对EEG多通道信号进行深度特征提取, 并提出一种基于广义正态分布优化的支持向量机(generalized normal distribution optimization based support vector machine, GNDO-SVM)情绪识别模型. 实验结果表明, 与基于遗传算法、粒子群算法和麻雀搜索算法优化的支持向量机模型相比, 所提出的GNDO-SVM模型具有更优的分类性能, 基于SAE深度特征的情感识别准确率达到了90.94%, 表明SAE能够有效地挖掘EEG信号不同通道间的深度相关性信息. 因此, 利用SAE深度特征结合GNDO-SVM模型可以有效地实现EEG信号的情绪识别.     

基于粒子群算法寻最优属性关联下的零样本语义自编码器

针对零样本图像分类构建共享属性层时造成的信息缺失问题,该文提出一种嵌入属性关联性的补偿方法.通过语义自编码器构建特征到属性的映射,然后以最大后验概率估计在类高斯模型构建的基础上实现零样本图像分类.为弥补SAE对属性关系学习的不足,引入加性因子与乘性因子对属性相关性进行嵌入,并利用粒子群算法搜寻最优的因子参数,实现属性相关性信息的补偿.实验结果表明采取相同映射方法的情况下,基于属性相关性嵌入的零样本图像分类在Pubfig数据集和OSR数据集上的分类效果较之其他方法得到了显著提升.     

一种基于SAE AS5643总线协议的驱动程序设计

结合SAE AS5643协议及其数据传输模型,设计出一种新的基于SAE AS5643总线协议的驱动程序及其架构图,并且开发了基于SAE AS5643系统的计算机客户端应用程序.该驱动程序使用异步流传输机制,用以保证传输的高可靠性.最后使用FireSpy通过工程实验证明了该驱动程序的可执行性.