一种用于分散化滤波的融合算法及其在组合导航系统中的应用

提出了一种局部估计相关时的最优融合算法，其特点是加权矩阵是对角阵，即全局估计中的各分量分别是局部估计中相应分量的线性组合，在此融合算法中局部估计是最优的，并且不存在由主滤波器向子滤波器的信息反馈，因此具有较好的客措性。给出了最优加权矩阵的存在性和唯一性的证明，并通过与现有类似融合算法的对比，表明该算法具有计算量小的优点。最后，以组合导航系统为例作了仿真计算，结果表明该融合算法是可行的。     

基于非结构化同位网格的SIMPLE算法

通过基于非结构化网格的有限体积法对二维稳态Navier—Stokes方程进行了数值求解。其中对流项采用延迟修正的二阶格式进行离散；扩散项的离散采用二阶中心差分格式；对于压力-速度耦合利用SIMPLE算法进行处理；计算节点的布置采用同位网格技术，界面流速通过动量插值确定。本文对方腔驱动流、倾斜腔驱动流和圆柱外部绕流问题进行了计算，讨论了非结构化同位网格有限体积法在实现SIMPLE算法时，迭代次数与欠松弛系数的关系、不同网格情况的收敛性、同结构化网格的对比以及流场尾迹结构。通过和以往结果比较可知，本文的方法是准确和可信的。     

结构拓扑优化设计的三角网格进化法

针对进化或拓扑优化方法的不足，提出了一种基于遗传算法的新型进化式拓扑优化方法－－三角网格进化法，该方法不仅能够同时进行拓扑，形状与截面变量优化设计，而且在优化过程中实现了退化和进化的统一，提高了优化效率。另外本文还首次对结构类型变量进行了优化计算，取得了有益的结果。最后几个数值算例证明了本方法的可行性和有效性。     

Real-Space Renormalization Estimates for Two-Phase Flow in Porous Media

We present a spatial renormalization group algorithm to handle immiscibletwo-phase flow in heterogeneous porous media. We call this algorithmFRACTAM-R, where FRACTAM is an acronym for Fast Renormalization Algorithmfor Correlated Transport in Anisotropic Media, and the R stands for relativepermeability. Originally, FRACTAM was an approximate iterative process thatreplaces the L × L lattice of grid blocks, representing the reservoir,by a (L/2) × (L/2) one. In fact, FRACTAM replaces the original L× L lattice by a hierarchical (fractal) lattice, in such a way thatfinding the solution of the two-phase flow equations becomes trivial. Thistriviality translates in practice into computer efficiency. For N=L ×L grid blocks we find that the computer time necessary to calculatefractional flow F(t) and pressure P(t) as a function of time scales as N^1.7 for FRACTAM-R. This should be contrasted with thecomputational time of a conventional grid simulator N^2.3. The solution we find in this way is an accurateapproximation to the direct solution of the original problem.     

计算具有区间参数结构的固有频率的优化方法

基于区间函数的单向包含性质，把具有区间非确定参数结构的固有频率所在区间范围问题 转化成两个全局优化问题，并采用一种实数编码遗传算法求取问题的全局解. 用一种能够求 得剪切型结构和弹簧质量系统特征值范围精确解的单调分析方法进行检验. 在一 些文献中，直接采用区间数运算法则和有限元法得到结构区间刚度阵和区间质量阵，并把关 于该区间刚度阵和区间质量阵的广义区间特征值问题的特征值区间作为待求的非确定性结构 的特征值所在的区间范围，该方法易于扩大问题的解域. 算例表明，可望得到结构 固有频率区间范围的准确解.     

条形平板中裂纹识别的遗传算法

研究条形平板中平行于上下边界的单一裂纹的识别问题。给出了当一声波入射这一条形平板时，裂纹参数与反射、透射系数之间的关系式，并应用遗传算法对裂纹进行了定量识别。计算结果表明本文给出的方法具有较好的识别精度。     

基于SIMPLER算法的多重网格方法研究

以二维方腔顶盖驱动流为模型,将多重网格方法和SIMPLER算法进行耦合,对不同雷诺数下多重网格加速SIMPLER算法和SIMPLER算法的计算效率进行了对比,数值计算表明:多重网格加速SIMPLER算法不仅能够解决SIMPLER算法不能准确模拟较高雷诺数流场的问题,而且其计算效率远远高于SIMPLER算法.本文也对松弛因子的选取、多重网格实现形式以及网格层数对多重网格加速SIMPLER算法的影响进行了研究,从而为多重网格加速SIMPLER算法的实施提供了计算技术.     

基于遗传算法的水轮机振动参数识别

通过对水轮机现场测试数据的频域分析，建立了测点位移响应的参数正模型。并基于遗传算法，建立了在时域内识别结构振动参数的数值方法，采用该方法对水轮机位移响应正模型的待定参数识别进行了研究。计算结果表明，所建立的基于遗传算法的参数识别方法具有良好的稳定性、抗观测噪声能力和较高的识别精度，能够反演出比较符合实际情况的位移响应振动模型。这就为荷载识别、结构优化和预报减震效果等后续工作提供了可靠的实施前提。     

A Non-linear Algorithm of Photoelastic Tomography for the Axisymmetric Problem

A non-linear algorithm of photoelastic tomography for the measurement of axisymmetric stress fields has been elaborated. It is free of any assumptions concerning the value of the birefringence or rotation of the principal stress axes along the light rays. The algorithm is based on the measurement of characteristic directions and phase retardation in two parallel sections of the test object. Stress components are presented in the form of power series along the radial coordinate. A differential evolution algorithm has been used for finding the stress field parameters, which fit the measurement data best. Application of the method is illustrated by residual stress measurement in a drinking glass.     

A new photoelastic model for studying fatigue crack closure

The photoelastic analysis of crack tip stress intensity factors has been historically developed for use on sharp notches in brittle materials that idealize the cracked structure. This approach, while useful, is not applicable to cases where residual effects of fatigue crack development (e.g., plasticity, surface roughness) affect the applied stress intensity range. A photoelastic model of these fatigue processes has been developed using polycarbonate, which is sufficiently ductile to allow the growth of a fatigue crack. The resultant stress field has been modeled mathematically using the stress potential function approach of Muskhelishvili to predict the stresses near a loaded but closed crack in an elastic body. The model was fitted to full-field photoelastic data using a combination of a generic algorithm and the downhill simplex method. The technique offers a significant advance in the ability to characterize the behavior of fatigue cracks with plasticity-induced closure, and hence to gain new insights into the associated mechanisms.