直扩系统中窄带干扰抑制技术

直扩系统中常常利用能量聚集度对窄带干扰进行定位,然后对干扰所在子带进行处理,从而达到抑制干扰的目的.当某次分解中左右节点中含有相近能量干扰时,将导致能量聚集度均匀而错误停止分解.为了解决这个问题,提出了一种利用多层分解对干扰进行定位的方法.当某节点的能量聚集度小于门限时,强迫子带算法继续分解并求取其子节点能量聚集度,对...     

Eolian origin of the Miocene loess-soil sequence at Qin’an, China: Evidence of quartz morphology and quartz grain-size

Recent study revealed that the so-called Neogene Gansu System in the western Loess Plateau contains loess-soil sequences of Miocene age. A most complete sequence （QA-I） covers a time interval from ～22 to ～.6.2 Ma BP, consisting of more than 230 visually definable paleosols interbedded with loess layers. A loess-soil sequence of late Miocene-Pliocene age, synchronous with the Hipparian Red Clay in the eastern Loess Plateau, was also recognized, and extends the Miocene sequences into the Pliocene, ～3.5 Ma BP. Recently, Miocene eolian deposits were also reported from the high terraces near Xining.     

面向网络管理知识获取的一种序列模式挖掘新算法

为了从反映网络状况和网络行为的海量历史数据中提取有用的网络管理知识,提出了一种用于获取网络管理知识的序列模式挖掘新算法--基于定位索引投影的序列模式挖掘算法.该算法采用模式增长、分而治之的思想,通过引入定位索引的概念,避免了对投影数据库的扫描,进一步减少了系统的开销.实验证明,利用该算法能够有效地发现网络告警事件序列模式.     

基于改进CLDNN的辐射源信号识别

传统辐射源信号识别方法往往需要人工提取特征,不仅对专业知识要求较高,而且人为选择的特征不能够保证适用于大多数类型信号的识别,识别精度和识别速度也不能兼顾.针对上述问题,将语音处理领域常用的深度学习模型——卷积长短时深度神经网络(convolutional long short-term deep neural netw...     

黄土公路隧道衬砌开裂分析

衬砌开裂是黄土公路隧道普遍存在的病害。为了预防和处治衬砌开裂,保证黄土隧道结构的长期稳定性,针对甘肃省榆中县新庄岭黄土公路隧道衬砌开裂的病害特征,运用数值仿真、实测等手段分析了病害产生的机理和原因。结果表明,黄土浸水导致土体强度降低、变形增大,从而增大了围岩塑性区范围和变形压力,恶化了衬砌结构受力状况,使衬砌结构拱顶内侧及拱脚外侧被拉裂,进而使拱脚内侧被压裂。根据隧道病害状况,提出了采用环氧树脂嵌补及凿槽嵌入钢拱架法处治衬砌开裂的建议,为分析和处治黄土公路隧道衬砌开裂提供了依据。     

试析大断面公路黄土隧道开挖施工坍塌的原因与预防

黄土地区开挖隧道易产生坍塌的施工事故,文章结合某工程工作实践,具体分析了在黄土隧道开挖中产生坍塌的原因,并对如何在施工中预防坍塌进行了探讨.     

广义高斯和分形序列及其M-J集研究

推广了Lakhtakia 和Berndt等的工作,分析了分形(自相似)序列的生成规则,给出了二次高斯和所生成的分形序列的标度及维数.利用逃逸时间算法,构造了广义高斯和的Mandelbrot-Julia集(M-J集),并从理论上分析了M-J集的周期性和结构特征.结果表明:M-J集由许多螺旋状的花束构成,这种结构在不同水平上嵌套出现,体现了明显的自相似分形特性;随指数值增大,M-J集中的精细花瓣结构增多并趋于复杂;J集在x轴方向上具有周期性.本研究成果有助于理解广义高斯和的动力学性?     

DMGSP:一种快速分布式全局序列模式挖掘算法

为了解决分布式环境下挖掘全局序列模式常产生过多候选序列,加大网络通信代价问题,提出了一种基于分布式环境下的快速挖掘全局序列模式算法--DMGSP.该算法将分布式环境下的各站点得到的局部序列模式压缩到一种语法序列树上, 避免了重复的序列前缀传输. 采用合并树中结点序列规则和项序扩展策略,对非频繁序列进行剪枝,有效地约简了候选序列,减少了网络传输量,从而快速生成全局序列模式.算法分析和实验结果表明,在大数据集环境下的DMGSP算法性能优越,能够有效地挖掘全局序列模式.     

动态数据系统(DDS)建模方法的改进

本文概述了原DDS建模方法的缺点,叙述了改进的途径,推导了改进的算法,给出了改进部分的详细框图。首先用长自回归拟合数据直到残差为白噪声,如AR模型阶不高,则认定为AR模型。否则,利用数据和残差,线性估计ARMA模型参数。增阶和降阶时,Yule—Wolker方程系数阵和模型参数都可递推求得。初估ARMA模型后,用非线性最小二乘法来优化参数。通过二个实例计算表明,经改进的方法可以节省66％的机时,效果是明显的。     

Mining rice new germplasm containing S5n gene by functional molecular marker and sequencing

Inter-subspecific hybrids between indica and japonica varieties yield strong biological heterosis, but it is difficult to utilize the hybrids directly in commercial production due to sterility of F1. A special gene S5^n may overcome the hybrids sterility, which is caused by the interaction between S5 loci. Recently, S5^n had been cloned, and it was revealed containing a large DNA deletion sequence that made the gene nonfunctional, compared to S5^i or S5^j. We designed a pair of primers flanking the deletion sequence of Ssn, and then applied to distinguish the varieties with S5^n or non-S5^n, convincing result suggested that the primers could be served as functional molecular marker to efficiently identify the new germplasm with S5^n, Using the functional marker, we surveyed 197 varieties from China National Micro-core Rice Collection, and found ten of which represented S5^n including following varieties： Haobuka, Sanbangqishiluo, Mubanggu, Xiaohonggu, Mowanggu＇neiza, Laozaogu, Fanhaopi, Feie＇nu02, Baoxie-7B, Teqingxuanhui. Among them, two varieties Sanbangqishiluo and Laozaogu was previously reported to contain S5^n gene. Further sequence analysis on the DNA sequence covering both sides of deletion in S5^n of the 10 varieties confirmed that the detected sequences in above varieties was identical with those of varieties containing S5^n, such as 02428 and Linglun. These results suggested that the gene in S5 locus of the ten varieties was also nonfunctional and it proved the presence of S5^n gene.