并行元胞单元法

在对元胞单元法的基本原理与特点进行简要介绍之后,探讨了元胞单元法的并行计算方法,给出了算法描述和解题步骤。以平面应力问题分析为例,进行了数值试验,给出了详细的计算步骤和结果。结果表明元胞单元法可形成一种高度并行的算法,适用于未来并行机的要求,有望在大型结构和纳米力学的大规模摸拟方面得到应用。     

元胞单元法

介绍一种力学计算方法——元胞单元法．它将结构的整体求解变成局域分析,通过力的局域间的不平衡传递达到最终的整体平衡．它对计算机容量要求低,在大规模计算方面具有良好的前景．同时,由于借用元胞自动机的思想,可形成一种高度并行的算法,适用于未来并行机的要求．是一种有发展潜力的数值方法,可望在大型结构和纳米力学的大规模摸拟方面得到应用．文中介绍了元胞单元法的基本原理,将其用于平面问题分析,进行了数值试验,给出了计算步骤和结果,并对其优点和存在的问题进行了讨论．     

基于元模型与聚类算法的设计空间减缩策略及工程应用

提出了一种基于元模型建模和聚类算法的设计空间减缩策略,能直接将设计空间减小到相对小的子空间。建议的方法分为三步:首先,进行低精度元模型建模,其次重新设计样本点并基于元模型求解样本点函数值和导数,应用最短距离层次聚类算法确定样本点聚类数与聚类中心,采用模糊c均值算法完成设计空间划分,生成设计空间子空间;最后,求解子空间目标函数的均值,确定保留子空间,并在保留子空间中对函数进行优化,达到目标函数全局最优解。测试函数和工程算例表明,该方法能够有效减小设计空间。     

二维弹性新型快速多极虚边界元的最小二乘法

在虚边界元最小二乘法的方程求解中采用新型的快速多极展开和广义极小残值法,提出了一种二维弹性新型快速多极虚边界元最小二乘法的求解思想。基于二维弹性问题原有的快速多极虚边界元最小二乘法的展开格式,通过引入对角化的概念,以更新展开传递格式;相对于原有快速多极算法,该方法可进一步提高计算效率且仍能保证具有较高的计算精度。数值算例说明了该方法的可行性、计算效率、计算精度均较高。     

四元数在惯性导航中的应用

在惯性导航系统中,用于姿态定位和控制的方法通常有三种:欧拉角法,方向余弦法,四元数法。比起前两种方法,四元数法具有明显的优点:方程不退化,数目较少,形式是线性的,特别在姿态变换算法中使用方便。     

h—层状自适应边界元方法的算法

给出了ｈ－层状自适应边界元方法的基本算法,并介绍了应用这一算法编写求平面弹性静力学问题的计算机程序Ｅｌｑｕａｈｈｂｅ。这处程序中包含一些新的内容：它的自适应细分过程通过一个新型的左指示因子加以驱动;整个细分过程的信息都压缩到两组“指针”数组中,对“指针”数组进行简单运算即可得到生成离散网格和装配系统矩阵所需的信息,因此较好地解决了数据管理中存储与导出的矛盾。数值算例证明这些方法是行之有效的。     

虚边界元最小二乘配点法

本方法是在虚边界上进行数值积分，在实边界上有限个点处满足给定问题边界条件的一种数值算法。在虚边界上积分是为能寻求到使原问题得到正确解的较好的分布虚体力；在实边界上配点，是依据加权残数法中超额配点，即最小二乘配点法的思想。文中给出了本文方法的基本思想，并给出了壳体的算例。由数值结果表明本文方法的计算精度是令人满意的。     

大型边界元方程组的并行直接分块求解算法

针对大型边界元方程组和网络微机机群环境提出了一种并行直接分块求解算法，算法基于分块高斯-若当消去法的原理，采用内外存交互技术，并行分块消去方法，节点超行的卷帘存储方案和并行环状循环逐次修正策略，增大了解题规模，提高了计算速度。算例计算结果表明该算法具有较高的并行加速比和并行效率，适用于大型问题的边界元法求解。     

大规模边界元分析的子域并行算法及其工程应用

为了扩大边界元多域分析的计算规模,对边界元子域并行算法进行了研究,针对不同的使用情况研制了不同的算法。算法引入分块技术以解决内存不足的矛盾,引入选主元或局部选主元技术以提高数值稳定性。特别是发展了多机一域算法,可以灵活解决不均匀子域划分的负载平衡问题。数值试验结果和实际工程应用都表明算法是有效的。     

8—21节点块体精化不协调元

本文基于文（２）提出的精化直接刚度法的变分原理，构造了８－２１节点元的精化不协调位移模式，并装入ＳＡＰ５程序中，该单元能通过分片试验数据结果表明：采用精化直接刚度法，可以有效地改装单元特性，使新单元在具原单元的简单、高效、可靠性好的同时，在精度，适应性等方面有大幅度的提高。     

三维弹塑性自然单元法算法实现

自然单元法是一种新兴的无网格数值计算方法,其实质是基于自然相邻插值(C∞)的伽辽金法。该方法计算精度与四边形或六面体单元有限元法相当,自然相邻插值函数比其他无网格法插值函数的计算速度快。由于自然相邻插值在凸域的边界上的相邻点之间是严格线性的,所以自然单元法在边界面的处理也相当简单。本文研究了在自然单元法中采用Von.Mises,Mohr-Coulomb和Drucker-Prager屈服准则解决三维弹塑性问题,并编制了相应计算程序,最后通过算例验证算法的正确性。     

三维非结构网格DSMC方法的实现及其应用

研究了三维非结构网格DSMC方法实现的过程。将Bird位置元方案中的子网格思想引入到非结构网格上来,只存储子网格的总体标识号,利用较少的计算网格提高了分子的分辨率与计算精度。提出了将体积元坐标搜索算法与交替数字二叉树搜索算法(ADT)相结合的方法来跟踪模拟分子在网格之间的迁移,使用ADT方法判别分子与物面是否作用,避免了分子表面反射的非确定论判据。利用Fortran 90的动态分配内存技术编制了通用计算程序。最后对高超声速过渡流域航天飞机头部外形绕流进行了数值模拟,数值结果初步验证了算法的可行性。     

Finite Element Error Analysis Approach for Three-Dimensional Incompressible Viscous Fluid Flow Analysis

In our previous research, the modified Galerkin method was proposed as one of the most efficient methods for the analyses of convection-diffusion problems and two-dimensional viscous fluid flow problems. In this modified Galerkin method, the inertia term is considered explicitly, so only the symmetrical matrixes appear. Then an artificial viscosity is introduced through an error analysis approach to improve its accuracy and stability. In this paper, we proposed a new finite element formulation for three-dimensional incompressible viscous fluid flow analysis. This formulation (‘MS’ algorithm and ‘MSR’ algorithm) is based on the modified Galerkin method coupled with the Semi-Implicit Method for Pressure-Linked Equations. The cubic cavity flow problems were investigated for the Reynolds number of 400, 1,000, 2,000 and 3,200 using non-uniform meshes. Finally, we confirmed the effectiveness of our proposed method through the comparison with other research works.     

部分输入未知时求解动力复合反演问题的补偿算法

将作者提出的全量补偿算法[8] 推广为一类适用于一般多自由度系统的时域补偿识别算法。该方法是基于最小二乘原则的一类迭代计算方法 ,对于结构上的部分作用力已知的情况 ,可用来同时识别结构的参数并反演输入。文中应用不同类型的结构分析实例说明了此方法适用于实际工程动力检测的可能性。     

空间刚架结构的广义逆力法

在结构计算中，根据算法中所采用的基本未知量的不同．结构分析方法可以分为力法、位移法和混合法。其中位移法由于适宜计算机处理而在结构计算领域得到了广泛的应用．经典力法相比之下应用就远不如位移法普遍，虽然力法本身在力学上有其独特的优势。广义逆矩阵做为一种较新的数学工具，自二十世纪五十年代诞生以来正日益表现出越来越旺盛的生命力。广义逆力法就是一种基于力法和广义逆矩阵理论的新的迭代算法。这种算法是一种完全适合计算机处理的力法方法。该算法的思路以及对于求解线弹性空间刚架结构问题的具体公式均在文中给出并给出了算例。从算例计算结果可以看到广义逆力法有着较好的计算效率和计算精度。该算法的提出为力法在计算机计算领域的应用开拓了新的发展空间。该算法在材料非线性问题和结构并行计算方面也有着较好的应用前景。     

A combined programming and iteration algorithm for finite element analysis of three-dimensional contact problems

Comparing with two-dimensional contact problems, three-dimensional frictional contact problems are more difficult to deal with, because of the unknown slip direction of the tangential force and enormous computing time. In order to overcome these difficulties, a combined PQP (Parametric Quadratic Programming) and iteration method is derived in this paper. The iteration algorithm, which alleviates the difficulty of unknown slip direction, is used along with the PQP method to cut down computing costs. Numerical example is given to demonstrate the validity of the present algorithm. The project supported by the Machinary and Electronics Ministry of China     

Effective numerical methods for elasto-plastic contact problems with friction

Several effective numerical methods for solving the elasto-plastic contact problems with friction are presented. First, a direct substitution method is employed to impose the contact constraint conditions on condensed finite element equations, thus resulting in a reduction by half in the dimension of final governing equations. Second, an algorithm composed of contact condition probes and elasto-plastic iterations is utilized to solve the governing equation, which distinguishes two kinds of nonlinearities, and makes the solution unique. In addition, Positive-Negative Sequence Modification Method is used to condense the finite element equations of each substructure and an analytical integration is introduced to determine the elasto-plastic status after each time step or each iteration, hence the computational efficiency is enhanced to a great extent. Finally, several test and practical examples are presented showing the validity and versatility of these methods and algorithms. The Project Supported by National Natural Science Foundation of China.     

A non‐iterative implicit algorithm for the solution of advection–diffusion equation on a sphere

A numerical algorithm for the solution of advection–diffusion equation on the surface of a sphere is suggested. The velocity field on a sphere is assumed to be known and non‐divergent. The discretization of advection–diffusion equation in space is carried out with the help of the finite volume method, and the Gauss theorem is applied to each grid cell. For the discretization in time, the symmetrized double‐cycle componentwise splitting method and the Crank–Nicolson scheme are used. The numerical scheme is of second order approximation in space and time, correctly describes the balance of mass of substance in the forced and dissipative discrete system and is unconditionally stable. In the absence of external forcing and dissipation, the total mass and L₂‐norm of solution of discrete system is conserved in time. The one‐dimensional periodic problems arising at splitting in the longitudinal direction are solved with Sherman–Morrison's formula and Thomas's algorithm. The one‐dimensional problems arising at splitting in the latitudinal direction are solved by the bordering method that requires a prior determination of the solution at the poles. The resulting linear systems have tridiagonal matrices and are solved by Thomas's algorithm. The suggested method is direct (without iterations) and rapid in realization. It can also be applied to linear and nonlinear diffusion problems, some elliptic problems and adjoint advection–diffusion problems on a sphere. Copyright © 2015 John Wiley & Sons, Ltd.     

An adaptive algorithm of precise integration for transient analysis

This paper presents an improved precise integration algorithm for transient analysis of heat transfer and some other problems. The original precise integration method is improved by means of the inverse accuracy analysis so that the parameterN, which has been taken as a constant and an independent parameter without consideration of the problems in the original method, can be generated automatically by the algorithm itself. Thus, the improved algorithm is adaptive and the accuracy of the algorithm is not dependent on the length of the time step in the integration process. It is shown that the numerical results obtained by the method proposed are more accurate than those obtained by the conventional time integration methods such as the difference method and others. Four examples are given to demonstrate the validity, accuracy and efficiency of the new method. Project supported by the National Natural Science Foundation of China (No. 19872016, 19872017), the National Key Basic Research Special Foundation (G1999032805) and the Foundation for University Key Teachers by the Ministry of Education of China.     

粘性流基于特征线的四阶Runge-Kutta有限元法

对于二维不可压缩粘性流，通过沿流线方向的坐标变换，推导了无对流项的二维N-S（Navier-Stokes）方程。采用四阶Runge-Kutta法对N-S方程进行时间离散，并沿流线进行Taylor展开，得到显式的时间离散格式，然后利用Galerkin法对其进行空间离散，得到了高精度的有限元算法。利用本文算法对方腔驱动流和圆柱绕流进行了数值计算，通过对时间步长、网格尺寸和流场区域的计算分析，进一步验证了本文算法相比经典CBS法在时间步长、收敛性、耗散性和计算精度方面更具有优势。