一种容错会议密钥分配方案设计与分析

基于Byzantine模型，提出了一种会议密钥分配方案．该方案通过将秘密信息分布于多个服务器中实现了安全和容错的密钥分配．在该方案中，即使实施密钥分发的服务器组中有部分被入侵或系统部件存在故障，系统的所有授权用户仍可以获得正确的会议密钥．对方案的分析表明，基于离散对数计算和大整数分解的困难性，任何攻击者都无法获得会议密钥的相关信息，同时入侵者亦不能阻止系统的授权用户得到正确的会议密钥。     

Difference schemes on non-uniform mesh and their application

High order accurate schemes are needed to simulate the multi-scale complex flow fields to get fine structures in simulation of the complex flows with large gradient of fluid parameters near the wall, and schemes on non-uniform mesh are desirable for many CFD (computational fluid dynamics) workers. The construction methods of difference approximations and several difference approximations on non-uniform mesh are presented. The accuracy of the methods and the influence of stretch ratio of the neighbor mesh increment on accuracy are discussed. Some comments on these methods are given, and comparison of the accuracy of the results obtained by schemes based on both non-uniform mesh and coordinate transformation is made, and some numerical examples with non-uniform mesh are presented.     

Numerical tests of efficiency of the retrospective time integration scheme in the self-memory model

A set of numerical tests was carried out to compare the retrospective time integral scheme in a self-memory model,whose dynamic kernel is the barotropical quasi-geostrophic model, with the ordinary centered difference scheme in the barotropical quasigeostrophic model. The Rossby-Haurwitz wave function was taken as the initial fields for both schemes. The results show that in comparison with the ordinary centered difference scheme, the retrospective time integral scheme reduces by 2 orders of magnitude the forecast error, and the forecast error increases very little with lengthening of the time-step. Therefore, the retrospective time integral scheme has advantages of improving the forecast accuracy, extending the predictable duration and reducing the computation amount.     

基于卷积神经网络的多天线Polar码联合解调-解码方案

为了验证运用神经网络进行信道解码的可行性,利用其提高短码长Polar码的译码准确率,提出了一种基于卷积神经网络(convolutional neural network,CNN)的多天线(multiple-input multiple-output)Polar码新颖联合解调-解码方案.搭建了一种包括4个卷积层,2个全连接层和1个输出层的卷积神经网络,采用最小均方误差作为损失函数,通过计算机生成了Polar码编码的多天线数据并对网络进行训练,使训练得到的神经网络能很好地提取出Polar码比特间的关系特征,从而拟合出Polar码译码函数.仿真结果表明,在相同信噪比条件下,基于CNN的Po-lar码联合解调-解码方案的误码率优于已有的基于全连接神经网络方案;所提方案在不同码率的仿真实验中的误码率皆优于基于全连接神经网络方案,损失曲线的收敛速度更快,显示了基于CNN的Polar码联合解调-解码方案具有更好的泛化能力和学习能力.     

自然崩落法放矿方案初步优化

放矿方案优化是放矿优化控制和放矿管理的基础。以数字矿山软件平台和VC++为工具,进行自然崩落法放矿方案优化研究。首先确定放矿方案构成要素与放矿方案优化准则;在数字矿山软件平台上构建放矿优化三维模型,确定放矿水平及放矿边界;自动确定放矿结构参数及布置放矿点,建立放矿点和放矿分层的品位计算数据搜索模型,通过品位估值确定放矿点及放矿分层的品位分布;建立基于矿山利润的放矿资源价值模型,以金属市场价格、生产成本及贫损指标为基础输入参数,计算各放矿分层的价值和利润,得到放矿点的最佳放矿高度、担负矿量和出矿截止品位,从而确定放矿优化方案,并为放矿计划编制和管理提供数据。     

有限体积法定价欧式跳扩散期权模型

考虑有限体积法求解Kou跳扩散期权定价模型.基于线性有限元空间,构造了向后Euler和Crank-Nicolson两种全离散有限体积格式,并结合简单高效的递推公式逼近方程中的积分项.理论分析表明所得的离散矩阵为M-矩阵.数值实验验证了方法的有效性.     

管道内运动火焰生成压力波及其特性的研究

从一维非定常守恒方程出发 ,本文用Richtmyer Lax Wendroff格式研究了压缩波、稀疏波、燃烧速率、火焰熄灭和管道长度对管内运动火焰生成压力波的影响 .结果表明 :相对于火焰面 ,同向压力波和异向稀疏波会加速火焰传播 ;异向压力波和同向稀疏波使火焰传播减速 ;增大燃烧速率使得和火焰面两侧的压力、温度增大 ;火焰熄灭时 ,火焰面变为接触间断面 ,同时产生稀疏波向火焰面两侧传播 .此后 ,接触间断面几乎处于滞止状态 ;增加管子长度会促使燃烧向爆轰转变 .     

Technology for long cable erection of a thousand meter scale cable stayed bridge

In the background of the construction of Sutong Yangtze River Bridge （ short as Sutong Bridge ）, the cable construction method and techniques of a thousand-meter scale cable-stayed bridge are introduced. Some key construction techniques, such as outspreading cable on deck, installing cable at pylon, pulling and fixing cable at the attachment with decks and cable PE sheath protection are discussed.     

General Improved KdV方程的三层加权平均线性差分格式

本文对广义Improved KdV方程的初边值问题进行了数值研究,提出了一个三层加权平均线性差分格式,分析了差分解的存在唯一性,证明了格式的二阶收敛性和稳定性.数值实验验证了差分格式的有效性.     

铝热自蔓延制备CuCr合金渣系的粘度测量及模型建立(Ⅰ)

根据Al2O3基渣系的结构特点将S.C.Du和S.Seetharaman等人的粘度数学模型应用于非二氧化硅体系,建立了针对铝热反应测量渣系粘度的模型·根据添加剂选择原则结合相图研究了CaO Al2O3和CaO Al2O3 MgO体系作为铝热自蔓延制备CuCr合金渣系组分的可行性·由实验测量值和模型计算值比较分析,该模型处理以Al2O3为基体的渣系是成功的·