Comparison of POD reduced order strategies for the nonlinear 2D shallow water equations

This paper introduces tensorial calculus techniques in the framework of POD to reduce the computational complexity of the reduced nonlinear terms. The resulting method, named tensorial POD, can be applied to polynomial nonlinearities of any degree p. Such nonlinear terms have an online complexity of , where k is the dimension of POD basis and therefore is independent of full space dimension. However, it is efficient only for quadratic nonlinear terms because for higher nonlinearities, POD model proves to be less time consuming once the POD basis dimension k is increased. Numerical experiments are carried out with a two‐dimensional SWE test problem to compare the performance of tensorial POD, POD, and POD/discrete empirical interpolation method (DEIM). Numerical results show that tensorial POD decreases by 76× the computational cost of the online stage of POD model for configurations using more than 300,000 model variables. The tensorial POD SWE model was only 2 to 8× slower than the POD/DEIM SWE model but the implementation effort is considerably increased. Tensorial calculus was again employed to construct a new algorithm allowing POD/DEIM SWE model to compute its offline stage faster than POD and tensorial POD approaches. Copyright © 2014 John Wiley & Sons, Ltd.     

On linear and nonlinear aspects of dynamic mode decomposition

The approximation of reduced linear evolution operator (propagator) via dynamic mode decomposition (DMD) is addressed for both linear and nonlinear events. The 2D unsteady supersonic underexpanded jet, impinging the flat plate in nonlinear oscillating mode, is used as the first test problem for both modes. Large memory savings for the propagator approximation are demonstrated. Corresponding prospects for the estimation of receptivity and singular vectors are discussed. The shallow water equations are used as the second large‐scale test problem. Excellent results are obtained for the proposed optimized DMD method of the shallow water equations when compared with recent POD‐based/discrete empirical interpolation‐based model reduction results in the literature. Copyright © 2016 John Wiley & Sons, Ltd.     

Non-intrusive reduced genetic algorithm for near-real time flow optimal control

Most genetic algorithms (GAs) used in the literature to solve control problems are time consuming and involve important storage memory requirements. In fact, the search in GAs is iteratively performed on a population of chromosomes (control parameters). As a result, the cost functional needs to be evaluated through solving the high fidelity model or by performing the experimental protocol for each chromosome and for many generations. To overcome this issue, a non-intrusive reduced real-coded genetic algorithm (RGA) for near real-time optimal control is designed. This algorithm uses precalculated parametrized solution snapshots stored in the POD (proper orthogonal decomposition) reduced form, to predict the solution snapshots for chromosomes over generations. The method used for this purpose is a economic reduced version of the Bi-CITSGM method (Bi-calibrated interpolation on the tangent space of the Grassmann manifold) designed specially for nonlinear parametrized solution snapshots interpolation. This approach is proposed in such a way to accelerate the usual Bi-CITSGM process by bringing this last to a significantly lower dimension. Thus, the whole optimization process by RGA can be performed in near real-time. The potential of RGA in terms of accuracy and central processing unit time is demonstrated on control problems of the flow past a cylinder and flow in a lid-driven cavity when the Reynolds number value varies.     

An optimizing implicit difference scheme based on proper orthogonal decomposition for the two‐dimensional unsaturated soil water flow equation

An optimizing reduced implicit difference scheme (IDS) based on singular value decomposition (SVD) and proper orthogonal decomposition (POD) for the two‐dimensional unsaturated soil water flow equation is presented. An ensemble of snapshots is compiled from the transient solutions derived from the usual IDS for a two‐dimensional unsaturated flow equation. Then, optimal orthogonal bases are reconstructed by implementing SVD and POD techniques for the ensemble of snapshots. Combining POD with a Galerkin projection approach, a new lower dimensional and highly accurate IDS for the two‐dimensional unsaturated flow equation is obtained. Error estimates between the true solution, the usual IDS solution, and the reduced IDS solution based on POD basis are derived. Finally, it is shown by means of a numerical example using the technology of local refined grids that the computational load is greatly diminished by using the reduced IDS. Also, the error between the POD approximate solution and the usual IDS solution is proved to be consistent with the derived theoretical results. Thus, both feasibility and efficiency of the POD method are validated. Copyright © 2011 John Wiley & Sons, Ltd.     

An optimizing reduced order FDS for the tropical Pacific Ocean reduced gravity model

Proper orthogonal decomposition (POD) and singular value decomposition (SVD) methods are used to study a finite difference discretization scheme (FDS) for the tropical Pacific Ocean reduced gravity model. Ensembles of data are compiled from transient solutions computed from the discrete equation system derived by FDS for the tropical Pacific Ocean reduced gravity model. The optimal orthogonal bases are used to reconstruct the elements of the ensemble with POD and SVD. Combining the above approach with a Galerkin projection procedure yields a new optimizing FDS model of lower dimensions and high accuracy for the tropical Pacific Ocean reduced gravity model. An error estimate of the new reduced order optimizing FDS model is then derived. Numerical examples are presented illustrating that the error between the POD approximate solution and the full FDS solution is consistent with previously obtained theoretical results, thus validating the feasibility and efficiency of POD method. Copyright © 2007 John Wiley & Sons, Ltd.     

基于特征正交分解的非定常气动力建模技术

采用特征正交分解(proper orthogonal decomposition, POD)方法, 建立了基于状态空间的非定常气动力降阶模型, 并耦合结构方程, 建立了降阶的气动弹性系统, 开展了颤振分析的初步研究, 计算效率提高了2~3个数量级. 具体过程是:首先获取全阶系统的频域快照构成关联矩阵, 通过对关联矩阵进行奇异值分解提取流场模态(或流场基), 对低能量模态截断形成降阶子空间, 并将其映射到全阶系统, 从而形成基于状态空间的降阶非定常气动力模型. 对气动弹性标模AGARD445.6进行算例验证, 证明了降阶方法正确, 可以提供高效、高精度的气动弹性分析.      

三维高频声波的矩阵压缩边界节点法模拟

本文采用边界节点法(Boundary Knot Method, BKM)求解三维高频声场．由于高频赫姆霍兹方程的解是振荡的，极大影响了数值求解的精确度，需要在计算区域增加离散点，这会增加计算量．同时对于大规模声学问题，依靠边界节点法形成的插值矩阵为满秩，导致计算量过高和存储量过大．所以本文采用矩阵压缩技术(Matrix Compression, MC)，在有效继承边界节点法高精确度的基础上减少计算内存需求和时间，从而提高计算效率．数值实验表明，MC-BKM 求解精度高、收敛速度快、计算时间少，在高频大规模声波问题中应用前景广泛．     

一种简单的局部离散速度统一气体动理论格式

统一气体动理论格式UGKS(Unified Gas-Kinetic Scheme)是一种适用于从连续流到自由分子流的全流域计算格式。在该格式中一般使用统一的离散速度空间。而在高速流动中,不同节点的分布函数往往差异很大。为了保证计算的精度,离散速度空间必须满足所有节点的需要,占用了大量的内存。采用局部的均匀离散速度空间,离散速度的范围随节点状态的变化而变化,从而降低了内存的需要,并通过引入背景网格避免了不同节点离散速度的插值。最后,通过两个一维算例对该方法进行了测试。测试结果显示,采用局部离散速度空间能够得到可靠的结果,并且在模拟高速流动时计算效率明显提高。     

PERTURBATION METHOD FOR REANALYSIS OF THE MATRIX SINGULAR VALUE DECOMPOSITION

The perturbation method for the reanalysis of the singular value decomposition(SVD)of general real matrices is presented in this paper.This is a simple but efficientreanalysis technique for the SVD,which is of great worth to enhance computationalefficiency of the iterative analysis problems that require matrix singular valuedecomposition repeatedly.The asymptotic estimate formulas for the singular values and thecorresponding left and right singular vectors up to second-order perturbation componentsare derived.At the end of the paper the way to extend the perturbation method to the case ofgeneral complex matrices is advanced.     

Parallel data-driven decomposition algorithm for large-scale datasets: with application to transitional boundary layers

Many fluid flows of engineering interest, though very complex in appearance, can be approximated by low-order models governed by a few modes, able to capture the dominant behavior (dynamics) of the system. This feature has fueled the development of various methodologies aimed at extracting dominant coherent structures from the flow. Some of the more general techniques are based on data-driven decompositions, most of which rely on performing a singular value decomposition (SVD) on a formulated snapshot (data) matrix. The amount of experimentally or numerically generated data expands as more detailed experimental measurements and increased computational resources become readily available. Consequently, the data matrix to be processed will consist of far more rows than columns, resulting in a so-called tall-and-skinny (TS) matrix. Ultimately, the SVD of such a TS data matrix can no longer be performed on a single processor, and parallel algorithms are necessary. The present study employs the parallel TSQR algorithm of (Demmel et al. in SIAM J Sci Comput 34(1):206–239, 2012), which is further used as a basis of the underlying parallel SVD. This algorithm is shown to scale well on machines with a large number of processors and, therefore, allows the decomposition of very large datasets. In addition, the simplicity of its implementation and the minimum required communication makes it suitable for integration in existing numerical solvers and data decomposition techniques. Examples that demonstrate the capabilities of highly parallel data decomposition algorithms include transitional processes in compressible boundary layers without and with induced flow separation.     

Proper application of a kind of matrix construction method in physical parameter identification of dynamic model

ＩｎｔｒｏｄｕｃｔｉｏｎＴｈｅｐｈｙｓｉｃｓｐａｒａｍｅｔｅｒｉｄｅｎｔｉｆｉｃａｔｉｏｎｐｒｏｂｌｅｍｓａｒｅｆｒｅｑｕｅｎｔｌｙｅｎｃｏｕｎｔｅｒｅｄｉｎｔｈｅｆｉｅｌｄｓｏｆａｖｉａｔｉｏｎ,ｓｐａｃｅ,ｍｅｃｈａｎｉｓｍａｎｄｃｉｖｉｌｅｎｇｉｎｅｅｒｉｎｇ .Ｆｏｒｅｘａｍｐｌｅ ,ｄｕｅｔｏｔｈｅｃｏｍｐｌｅｘｉｔｙｏｆｍａｔｅｒｉａｌｐｒｏｐｅｒｔｉｅｓｉｎｃｉｖｉｌｅｎｇｉｎｅｅｒｉｎｇ ,ｔｈｅｐｈｙｓｉｃｓｐａｒａｍｅｔｅｒｓｕｓｕａｌｌｙａｒｅｉｄｅｎｔｉｆｉｅｄｂｙｔｅｓｔｄａｔ…     

基于分区径向基函数配点法的大变形分析

无网格法因为不需要划分网格, 可以避免网格畸变问题,使得其广泛应用于大变形和一些复杂问题. 径向基函数配点法是一种典型的强形式无网格法,这种方法具有完全不需要任何网格、求解过程简单、精度高、收敛性好以及易于扩展到高维空间等优点,但是由于其采用全域的形函数, 在求解高梯度问题时 存在精度较低和无法很好地反应局部特性的缺点. 针对这个问题,本文引入分区径向基函数配点法来求解局部存在高梯度的大变形问题. 基于完全拉格朗日格式,采用牛顿迭代法建立了分区径向基函数配点法在大变形分析中的增量求解模式.这种方法将求解域根据其几何特点划分成若干个子域, 在子域内构建径向基函数插值, 在界面上施加所有的界面连续条件,构建分块稀疏矩阵统一求解. 该方法仍然保持超收敛性, 且将原来的满阵转化成了稀疏矩阵, 降低了存储空间,提高了计算效率. 相比较于传统的径向基函数配点法和有限元法, 这种方法能够更好地反应局部特性和求解高梯度问题.数值分析表明该方法能够有效求解局部存在高梯度的大变形问题.      

混凝土细观力学分析程序中的快速算法与并行算法设计

针对一套混凝土细观力学分析程序,在分析其计算方法与计算效率的不足之后,提出了采用稀疏矩阵与稀疏向量技术来高效实现有限元刚度矩阵装配过程的算法,并采用双门槛不完全Cholesky分解预条件技术与CG法相结合来高效地求解稀疏线性方程组。之后,从整体上提出了一个将有限单元分布与未知量分布有机结合的并行算法设计方案,并分别针对刚度矩阵装配、双门槛不完全Cholesky分解、稀疏矩阵与稠密向量相乘、稀疏向量相加等核心算法,进行了相应的并行算法设计。最后,在由每节点2 CPU的8个Intel Xeon节点采用千兆以太网连成的机群上,针对两个混凝土数值试样进行了数值实验,第一个试样含44117个网格点与53200个有限单元,第二个试样含71013个网格点与78800个有限单元;对第一个试样,原串行程序进行全程567次加载计算需要984.83小时约41天,采用文中串行算法后,模拟时间减少到22531秒约6.26小时,采用并行算法在16个CPU上的模拟时间进一步降为3860秒约1.07小时。对第二个试样,原串行程序进行全程94次加载计算需要467.19小时约19.5天,采用文中串行算法后,模拟时间减少到11453秒约3.18小时,采用并行算法在16个CPU上的模拟时间进一步降为1704秒约28.4分钟。串行算法的改进与并行算法的设计大大缩短了计算时间,对加快混凝土力学性能的分析研究具有重要意义。     

基于控制响应的时变系统模态参数辨识的改进子空间方法

提出了一种基于系统控制信号激发的响应数据来辨识时变系统模态参数的改进子空间方法。该方法以系统控制响应信号建立系统的状态空间输出方程并构造了一个广义Hankel矩阵,通过对该矩阵做奇异值分解(SVD),用广义能观阵的估计代替输出矩阵,然后利用奇异值矩阵的正交性,有效地降低了噪声敏感性和计算量,从而容易地辨识出等效状态下的系统矩阵,最后采用转换矩阵辨识出时变系统的模态参数。通过理论分析、仿真和实验,讨论了不同信噪比对辨识结果的影响,验证了该方法的有效性。     

An indirect shooting method based on the POD/DEIM technique for distributed optimal control of the wave equation

This paper presents a fast numerical method, based on the indirect shooting method and Proper Orthogonal Decomposition (POD) technique, for solving distributed optimal control of the wave equation. To solve this problem, we consider the first‐order optimality conditions and then by using finite element spatial discretization and shooting strategy, the solution of the optimality conditions is reduced to the solution of a series of initial value problems (IVPs). Generally, these IVPs are high‐order and thus their solution is time‐consuming. To overcome this drawback, we present a POD indirect shooting method, which uses the POD technique to approximate IVPs with smaller ones and faster run times. Moreover, in the presence of the nonlinear term, to reduce the order of the nonlinear calculations, a discrete empirical interpolation method (DEIM) is applied and a POD/DEIM indirect shooting method is developed. We investigate the performance and accuracy of the proposed methods by means of 4 numerical experiments. We show that the presented POD and POD/DEIM indirect shooting methods dramatically reduce the CPU time compared to the full indirect shooting method, whereas there is no significant difference between the accuracy of the reduced and full indirect shooting methods.     

A SINGULAR VALUE DECOMPOSITION BASED TRUNCATION ALGORITHM IN SOLVING THE STRUCTURAL DAMAGE EQUATIONS

The structural damage identification through modal data often leads to solving a set of linear equations. Special numerical treatment is sometimes required for an accurate and stable solution owing to the ill conditioning of the equations. Based on the singular value decomposition (SVD) of the coefficient matrix, an error based truncation algorithm is proposed in this paper. By rejection of selected small singular values, the influence of noise can be reduced. A simply-supported beam is used as a simulation example to compare the results to other methods.Illustrative numerical examples demonstrate the good efficiency and stability of the algorithm in the nondestructive identification of structural damage through modal data.     

Computations of modes I and II stress intensity factors of sharp notched plates under in-plane shear and bending loading by the fractal-like finite element method

The fractal-like finite element method (FFEM) is used to compute the stress intensity factors (SIFs) for different configurations of cracked/notched plates subject to in-plane shear and bending loading conditions. In the FFEM, the large number of unknown variables in the singular region around a notch tip is reduced to a small set of generalised co-ordinates by performing a fractal transformation using global interpolation functions. The use of exact analytical solutions of the displacement field around a notch tip as the global interpolation functions reduces the computational cost significantly and neither post-processing technique to extract SIFs nor special singular elements to model the singular region are required. The results of numerical examples of various configurations of cracked/notched plates are presented and validated via published data. Also, new results for cracked/notched plate problems are presented. These results demonstrate the accuracy and efficiency of the FFEM to compute the SIFs for notch problems under in-plane shear and bending loading conditions.     

一种模拟大规模高频声场的双层奇异边界法

大规模高频声场的数值模拟是一项非常有计算挑战性的课题. 为了解决传统边界型离散方法由于全局支撑的满阵限制, 不易应用于大规模高频声场模拟的计算瓶颈, 本文提出了一种用于模拟大规模高频声场的双层奇异边界法. 在该方法中, 通过引入双层结构, 细网格上的全局支撑的满阵被转化为局部支撑的大规模稀疏矩阵, 传统奇异边界法模拟大规模问题时所面临的高计算量以及过度存储需求遂得以解决. 其次, 双层奇异边界法仅通过粗网格评估远场作用, 且独立于特定的插值核函数. 相较于快速多级方法, 该方法具有更强的适应性和灵活性, 且多层结构使该方法具有一定的预调节作用, 非常适合求解具有大规模、高秩、高条件数特点的高频波矩阵. 在其后的散射球模型算例中, 双层奇异边界法配置10万个节点, 成功模拟了无量纲波数高达160的声散射问题. 在对于人头模型的声散射特性分析中, 双层奇异边界法比COMSOL软件计算速度快了约78.13\mathrm{\%}. 当配置8万个节点时, 双层奇异边界法成功模拟了频率高达25 kHz 的工况, 该频率已远远超出了人耳的听力极限.      

基于SVD方法的lNS传递对准的可观测性能分析

在惯导系统一般的误差动态方程和速度观测方程的基础上，建立了姿态传递对准所必需的弹、舰相对姿态误差方程和观测方程。介绍了基于动态系统可观测性矩阵奇异值分解的状态变量可观测度的分析方法。用奇异值分解的方法，对同时采用速度和姿态传递的INS对准模型，分析了系统变量的可观测性和可观测度，为对准方程的可观测性结构分解和误差估计性能的改善提供了必要的基础。     

Singularity-free augmented Lagrangian algorithms for constrained multibody dynamics

After a general review of the methods currently available for the dynamics of constrained multibody systems in the context of numerical efficiency and ability to solve the differential equations of motion in singular positions, we examine the acceleration based augmented Lagrangian formulations, and propose a new one for holonomic and non-holonomic systems that is based on the canonical equations of Hamilton. This new one proves to be more stable and accurate that the acceleration based counterpart under repetitive singular positions. The proposed algorithms are numerically efficient, can use standard conditionally stable numerical integrators and do not fail in singular positions, as the classical formulations do. The reason for the numerical efficiency and better behavior under singularities relies on the fact that the leading matrix of the resultant system of ODEs is sparse, symmetric, positive definite, and its rank is independent of that of the Jacobian of the constraint equations. The latter fact makes the proposed method particularly suitable for singular configurations.