基于稀疏强特征提取的三维地震数据完备方法

随着复杂储层地震资料特征筛选的机器学习技术的进步,如何有效地对参与地震属性优选和储层反演的地震样本进行采集和分析,成为目前智能地震预测领域的一个研究热点。目前的方法多着重于模型分类算法的改进,在标签的制作和采集方面不仅耗费大量时间进行人工标注,还存在标签不平衡情况下类内可靠性、类间平衡性不强等问题。为此,提出基于稀疏强特征提取的三维地震数据完备方法。首先,基于多数决原则的样本分割(Sample Segmentation Based on Majority Rule, SSMR)寻迹多尺度、多标签三维地震样本,进行采集、自动标注;然后,改进标签洗牌平衡方法(Improved Label Shuffling Balance Method, ILSB),通过“2+1”的样本增广平衡策略进行数据完备处理,改善样本采样不平衡性导致的模型训练偏向性;最后,利用基于最小L₁范数稀疏表示对奇异值分解结果进行强特征提取(Minimum L₁-norm Based Sparse Representation for Feature Extraction, L     

Recent progress of CERN RD50 Collaboration

The objective of the CERN RD50 Collaboration is to develop radiation hard semiconductor detectors for very high luminosity colliders, in particular, for the upgrade of the large hadron collider （LHC） which itself is scheduled to be operational in 2007. The approach of the RD50 has two major research lines, material engineering and device engineering. These are further subdivided into projects covering defect characterization and engineering, new detector materials, detector characterization, new detector structures and full detector systems. Presently, 264 members from 53 institutes are actively participating in the RD50 Collaboration. Detectors made of defect engineered substrates, e.g. high resistivity magnetic Czochralski （MCz-Si）, epitaxial silicon （Epi-Si） on Czochralski silicon （Cz-Si） substrate, intentionally thermal donor （TD） compensated p-type MCz-Si and oxygen enriched （DOFZ） silicon, have been demonstrated by the RD50 Collaboration. An overview and highlights of the results of these defect engineering techniques were given in this report.     

Thermophysical properties of a Ti–44%Al–8%Nb–1%B alloy in the solid and molten states

The families of titanium aluminide intermetallic alloys have attractive high temperature mechanical properties which make them potential candidate materials for a wide range of applications, particularly in the aeronautic and automobile sectors. The development of appropriate manufacturing techniques is an essential stage in the engineering exploitation of these materials, e.g., Induction Skull Melting is one of the techniques which needs to be optimised for the casting of titanium aluminides. Research is underway to develop a computer model of this process but data are required for the key thermophysical properties. Pulse-heating techniques have been used to measure properties for the Ti–44Al–8Nb–1B system. Rectangular samples have been prepared and are resistively heated as part of a fast capacitor discharge circuit. Time-resolved measurements with sub-μs resolution of currents through the specimen were made with a Pearson probe current monitor using the induction principle. Voltages across the specimen were determined with knife-edge contacts and voltage dividers, and radiance temperatures of the sample were measured with a pyrometer. These measurements allow the calculation of specific heat and dependencies between enthalpy, electrical resistivity and temperature of the alloy up into the liquid phase. Data for thermal diffusivity have been obtained by using the Wiedeman–Franz relation. The results are compared with those obtained using DSC and the four-probe method to measure the temperature dependence of the resistivity.     

谈谈水泥土挡墙的合理计算

阐述了水泥土挡墙在工程基坑支护中的作用,并结合工程实例,对水泥土挡墙的合理计算模型进行了探讨。     

High-temperature oxidation of SiC in a glow-discharge oxygen plasma

The oxidation kinetics and structure of the oxide scales formed on high-density SiC were studied in molecular oxygen at 740 Torr and in a glow-discharge oxygen plasma at 0.1 Torr at temperatures of 1000, 1100, and 1200°C. The monatomic oxygen formed by the glow discharge markedly increased the reaction rate and the vaporization of some of the oxidation products. The marked differences in kinetics suggest that the rate-controlling step during oxidation in molecular oxygen is the dissociation of adsorbed diatomic oxygen to the monatomic species. Films formed in molecular oxygen were mostly amorphous SiO₂ with small inclusions of SiC and graphite, whereas films formed in dissociated oxygen were primarily amorphous SiO₂ containing SiO, S₂O₃, and the coesite form of SiO₂.     

初探混联法生产氧化铝的工艺优化

在目前的生产条件下。通过大量的混联法物料衡算，定量地研究了供矿铝硅比、熟料铝硅比和拜年法赤泥铝硅比对混联法的制造成本、能耗和拜年法与烧结法的生产比例的影响规律，提出我国混联法生产进一步优化的方向和途径。并开发了研究和评价的计算软件。为混联法的生产和设计的进一步优化奠定基础。     

云计算身份认证模型研究

云计算是在继承和融合众多技术基础上的一个突破性创新,已成为当前应用和研究的重点与热点。其中,云用户与云服务之间以及云平台中不同系统之间的身份认证与资源授权是确保云计算安全性的前提。在简要介绍云计算信息基础架构的基础上,针对云计算统一身份认证的特点和要求,综合分析了SAML2.0、OAuth2.0和Open ID2.0等技术规范的功能特点,提出了一种开放标准的云计算身份认证模型,为云计算中逻辑安全域的形成与管理提供了参考。     

基于NiosII的多通道HART智能通信系统设计

HART通信方式具有抗干扰、传输距离远等优点.针对传统HART通信模块体积大、功耗高、不利于集成的缺点,设计了一种基于NiosⅡ软核CPU的多通道HART智能通信系统.采用NiosⅡ软核CPU为处理核心,实现8路4～ 20 mA电流输出和8路HART通信,每个4～ 20 mA电流输出配置一个HART信号调制解调单元;在NiosⅡ软核CPU上挂载μC/OS-Ⅱ操作系统并实现8路数据链路的软件设计;设计采用DC-DC电源为D/A转换芯片提供动态电压,降低了系统功耗.设计的智能通信系统具有功耗低、集成度高的优点,在工业过程控制等恶劣环境下应用前景广泛.     

基于曲率匹配和递归排序的自适应排样算法

为提高盘状毛坯的使用率,提出一种口腔修复加工中模型边界在盘状毛坯中的排样算法,主要包括边界匹配、多边形定位及递归排序.首先基于协方差矩阵及矩阵SVD分解算法对模型多边形进行分段,并采用等弧长曲线采样曲率匹配确定待匹配模型边界;然后依毛坯边界角度分布及沿圆周排样的思想确定模型轮廓的旋转和平移定位;最后提出一种基于包络率的递归排序算法,对在排样过程中发现的大孔洞可动态地调整排样顺序.实验结果表明,该算法可以处理不规则模型边界在任意形状毛坯中的排样,能有效地降低加工成本.