基于矩阵指数函数Laguerre多项式展开的模型降阶方法

在经典平衡截断模型降阶方法的基础上, 提出一种基于矩阵指数函数Laguerre多项式展开的模型降阶方法. 该方法首先利用矩阵指数函数的Laguerre多项式展开, 给出控制系统可控Gram矩阵和可观Gram矩阵的近似低秩分解, 然后构造正交投影变换得到近似平衡系统, 进而通过截断较小的Hankel奇异值对应的状态得到降阶系统. 该方法计算高效, 且具有一定的自适应性. 最后, 通过数值算例验证算法的有效性.     

N阶含参数伸展矩阵的构造与实例

利用特例归纳,构造几类n阶含参数的伸展矩阵.这些伸展矩阵满足以下特点:1)带参数的任意n阶方阵;2)任意n阶伸展矩阵的行列式是确定的常数.基于相似矩阵的性质,得到几个重要构造定理,并且给出了相应的实例演算.     

MISO的离散动力系统的最小二乘模型降阶方法

利用最小二乘原理,提出一个基于SVD-Krylov的模型降阶方法,方法兼顾基于SVD模型降阶方法的理论性质和基于Krylov模型降阶方法的有效计算,使得到的降阶系统既能匹配原系统的前r阶模,又能够保持系统的稳定性.利用对称矩阵特征值的极小极大原理,给出了保持系统稳定性的一个新的证明方法,与已有的方法相比,提出的理论证明方法更为简洁.对于离散系统,方法除了能匹配原模型的前r个Markov参数,还可将其推广到任意点处模匹配.数值例子也证明了方法的有效性.     

基于Mori-Zwanzig格式和偏最小二乘的非线性模型降阶

本征正交分解及Galerkin投影是解决复杂非线性系统模型降阶问题常用的方法.然而,该方法在构造降阶系统过程中只截取基函数的部分模态,这通常会使得降阶系统不准确.针对该问题,提出了对降阶系统误差进行快速校正的方法.首先应用Mori-Zwanzig格式对降阶系统的误差进行分析,理论上得到误差模型的形式和有效预测变量.再通过偏最小二乘方法构造预测变量和系统误差的多元回归模型,建立误差预测模型.将所构造的误差预测模型直接嵌入到原降阶系统,得到新的降阶系统在形式上等价于对原模型的右端采用Petrov-Galerkin投影.最后给出了新的降阶系统的误差估计.数值结果进一步说明了所提方法能有效地提高降阶系统的稳定性和准确性,且具有较高计算效率.     

二阶系统基于奇异值的保结构模型降阶方法（英文）

本文提出了一种二阶系统基于奇异值的保结构模型降阶方法.该方法通过脉冲响应的移位勒让德多项式展开系数直接计算得到二阶系统的可控和可观格拉姆矩阵的低秩因子.然后基于二阶系统的奇异值结构,构造相应的平衡系统,进而通过截断较小奇异值对应的状态得到保结构的降阶系统.最后,通过两个数值实验展示了所提方法的有效性.     

径向基函数参数化翼型的气动力降阶模型优化

基于小扰动和弱非线性假设,提出了一种基于气动力降阶模型和径向基函数参数化的翼型优化方法.其主要方法是用径向基函数参数化翼型扰动;通过CFD辨识参数扰动对翼型气动力影响的降阶模型核函数;基于叠加法建立了参数变化对翼型气动力影响的降阶模型;最后基于该气动力降阶模型计算并优化翼型升阻特性.NACA0012翼型优化的结果表明基于气动力降阶模型的优化方法是可行的,可以极大地提高翼型优化速度.     

一种改进的线性系统的有限时间平衡截断（英文）

本文提出一种改进的线性系统的有限时间平衡截断方法.该方法首先利用Shifted Legendre多项式对线性系统的有限时间可控Gram矩阵和可观Gram矩阵进行近似低秩分解,其中根据正交多项式与幂级数之间的关系,该近似低秩分解因子可以通过简单的递推公式得到,然后构造正交投影变换得到近似平衡系统,进而通过截断较小的Hankel奇异值对应的状态得到降阶系统.此外,本文还简要讨论了该降阶模型的稳定性.最后,通过数值算例验证了算法的有效性.     

基于等价输入扰动的空间桁架振动控制的研究(英文)

基于等价输入扰动(EID)的方法,通过对抑制外界未知扰动(匹配的或者非匹配的),实现了对空间桁架的振动控制.首先利用有限元分析(FEA),计算空间桁架的质量矩阵,阻尼矩阵,刚度矩阵及输入矩阵,进而建立空间桁架的模型.为了便于分析和设计,利用模态降阶的方法对模型进行降阶处理.在降阶的模型的基础上,设计等价输入扰动观测器观测外部未知扰动.基于观测的结果,进一步地设计了空间桁架的振动控制方法.最后给出数值仿真,以证实所提方法的有效性.     

矩阵的逆及秩的降阶方法

降阶方法是处理矩阵问题的最核心的思想方法之一.从分块矩阵■出发,利用降阶的思想,讨论了该矩阵的逆与秩的计算,并给出该降阶公式的各种变形以及在解题中的应用.     

非线性基准建筑物振动的主动控制

非线性基准建筑物的振动方程属于非仿射系统,目前的非线性模型降阶方法不能采用．而直接采用非线性控制策略所设计的控制器阶数较高,难以用于实际场合．为此,开发了一种适合于非线性建筑结构的新的振动主动控制方法,该方法思路是识别线性化的结构模型,进而根据力作用原理把控制力施加到所识别的结构模型上．该方法所建模型可以通过经验Grammian矩阵进行平衡降阶,所以具有较好的实用性．最后给出了3层基准结构的计算实例,其结果表明所提出的方法对土木工程结构是可行的．     

Autoregression Model of Time Series with Matrix Cross-Section Data北大核心CSCD

矩阵型截面数据时间序列的优点在于可以同时刻画多个对象的多个属性.本文重点研究了矩阵型截面数据时间序列的自回归模型,给出了该模型的参数估计、模型识别、白噪声检验三个方面的理论结果.最后再利用矩阵型截面数据时间序列自回归模型,对两支银行股的日收益率序列和日成交量变化率序列进行建模分析.     

Derivation of the Double Porosity Model of a Compressible Miscible Displacement in Naturally Fractured Reservoirs

We derive a double porosity model for a displacement of one compressible miscible fluid by another in a naturally fractured reservoir. The microscopic model consists of the usual equations describing Darcy flow in a reservoir except that the porosity and the permeability coefficients are highly discontinuous. The viscosity is assumed to be constant. Over the matrix domain, the coefficients are scaled by a parameter ? representing the size of the matrix blocks. This scaling preserves the physics of the flow in the matrix as ? tends to zero. Using homogenization theory, we derive rigorously the corresponding double porosity model. To this purpose, we mainly use the concept of two-scale convergence. The less permeable part of the rock then contributes as nonlinear memory terms. To specify them in spite of the strong nonlinearities and of the coupling, we then use some appropriate dilation operator.     

基于遗传算法的LuGre轮胎模型参数辨识研究

LuGre轮胎模型是一种动态轮胎摩擦力模型,该模型能够精确描述轮胎摩擦环节的动态特性,但由其高度非线性使得参数辨识非常困难.针对LuGre轮胎模型,提出一种基于遗传算法的模型参数两步辨识方法.首先由PD控制辨识出静态参数;然后由PID控制辨识出动态参数.在每一步辨识中,均采用遗传算法作为优化工具,从而避免了采用拟和辨识方法中误差较大,试验条件难以控制的缺点.该算法仅仅使用轮胎转速数据,而转速传感器是汽车防滑刹车控制系统(ABS)的基本组成部分,因此该算法可以与ABS结合工作,低成本的实现LuGre轮胎模型参数辨识.     

On-line identification of language measure parameters for discrete-event supervisory control

The discrete-event dynamic behavior of physical plants is often represented by regular languages that can be realized as deterministic finite state automata (DFSA). The concept and construction of signed real measures of regular languages have been recently reported in literature. Major applications of the language measure are: quantitative evaluation of the discrete-event dynamic behavior of unsupervised and supervised plants; and analysis and synthesis of optimal supervisory control algorithms in the discrete-event setting. This paper formulates and experimentally validates an on-line procedure for identification of the language measure parameters based on a DFSA model of the physical plant. The recursive algorithm of this identification procedure relies on observed simulation and/or experimental data. Efficacy of the parameter identification procedure is demonstrated on the test bed of a mobile robotic system, whose dynamic behavior is modelled as a DFSA for discrete-event supervisory control.     

热传导(对流-扩散)方程源项识别的粒子群优化算法

提出了利用粒子群优化(PSO)算法反演热传导方程与对流-扩散方程源项的一种新方法,在已有文献方法的基础上,求解出这两类方程正问题的解析解,再把源项识别问题转化为最优化问题,结合粒子群优化算法寻优求解.通过数值模拟与统计检验,结果表明,此方法可快速有效地实现热传导方程与对流-扩散方程源项的识别,并可推广应用到其它数学物理方程的源项或参数的反演识别.     

Model recovery for Hammerstein systems using the auxiliary model based orthogonal matching pursuit method

This article investigates parameter and order identification of a block-oriented Hammerstein system by using the orthogonal matching pursuit method in the compressive sensing theory which deals with how to recover a sparse signal in a known basis with a linear measurement model and a small set of linear measurements. The idea is to parameterize the Hammerstein system into the linear measurement model containing a measurement matrix with some unknown variables and a sparse parameter vector by using the key variable separation principle, then an auxiliary model based orthogonal matching pursuit algorithm is presented to recover the sparse vector.The standard orthogonal matching pursuit algorithm with a known measurement matrix is a popular recovery strategy by picking the supporting basis and the corresponding non-zero element of a sparse signal in a greedy fashion. In contrast to this, the auxiliary model based orthogonal matching pursuit algorithm has unknown variables in the measurement matrix. For a K-sparse signal, the standard orthogonal matching pursuit algorithm takes a fixed number of K stages to pick K columns (atoms) in the measurement matrix, while the auxiliary model based orthogonal matching pursuit algorithm takes steps larger than K to pick K atoms in the measurement matrix with the process of picking and deleting atoms, due to the gradually accurate estimates of the unknown variables step by step.The auxiliary model based orthogonal matching pursuit algorithm can simultaneously identify parameters and orders of the Hammerstein system, and has a high efficient identification performance.     

Construction of kinetic models for metabolic reaction networks: Lessons learned in analysing short-term stimulus response data

Construction of dynamic models of large-scale metabolic networks is one of the central issues in the engineering of living cells. However, construction of such models is often hampered by a number of challenges, for example, data availability, compartmentalization and parameter identification coupled with design of in vivo perturbations. As a solution to the latter, short-term perturbation experiments are proposed and are proven to be a useful experimental method to obtain insights into the in vivo kinetic properties of the metabolic pathways.

The aim of this work is to construct a kinetic model using the available experimental data obtained by short-term perturbation experiments, where the steady state of a glucose-limited anaerobic chemostat culture of Saccharomyces cerevisiae was perturbed. In constructing the model, we first determined the steady-state flux distribution using the data before the glucose pulse and the known stoichiometry. For the rate expressions, we used approximative linlog kinetics, which allows the enzyme–metabolite kinetic interactions to be represented by an elasticity matrix. We performed a priori model reduction based on timescale analysis and parameter identifiability analysis allowing the information content of the experimental data to be assessed. The final values of the elasticities are estimated by fitting the model to the available short-term kinetic response data.

The final model consists of 16 metabolites and 14 reactions. With 25 parameters, the model adequately describes the short-term response of the cells to the glucose perturbation, pointing to the fact that the assumed kinetic interactions in the model are sufficient to account for the observed response.     

Wave parameter identification problem for ocean test structure data,part 2, discrete formulation

This paper deals with the solution of the wave parameter identification problem for ocean test structure data. A discrete formulation is assumed. An ocean test structure is considered, and wave elevation and velocities are assumed to be measured with a number of sensors. Within the frame of linear wave theory, a Fourier series model is chosen for the wave elevation and velocities. Then, the following problem is posed: Find the amplitudes of the various wave components of specified frequency and direction, so that the assumed model of wave elevation and velocities provides the best fit to the measured data. Here, the term best fit is employed in the least-square sense over a given time interval.At each time instant, the wave representation involves four indexes (frequency, direction, instrument, time); hence, four-dimensional arrays are required. This formal difficulty can be avoided by switching to an alternative representation involving only two indexes (frequency-direction, instrument-time); hence, standard vector-matrix notation can be used. Within this frame, optimality conditions are derived for the amplitudes of the assumed wave model.A characteristic of the wave parameter identification problem is that the condition number of the system matrix can be large. Therefore, the numerical solution is not an easy task and special procedures must be employed. Specifically, Gaussian elimination is avoided and advantageous use is made of the Householder transformation, in the light of the least-square nature of the problem and the discretized approach to the problem.Numerical results are presented. The effect of various system parameters (number of frequencies, number of directions, sampling time, number of sensors, and location of sensors) is investigated in connection with global or strong accuracy, local or weak accuracy, integral accuracy, and condition number of the system matrix.From the numerical experiments, it appears that the wave parameter identification problem has a unique solution if the number of directions is smaller than or equal to the number of sensors; it has an infinite number of solutions otherwise. In the case where a unique solution exists, the condition number of the system matrix increases as the size of the system increases, and this has a detrimental effect on the accuracy. However, the accuracy can be improved by proper selection of the sampling time and by proper choice of the number and location of the sensors.Generally speaking, the computations done for the discrete case exhibit better accuracy than the computations done for the continuous case (Ref. 5). This improved accuracy is a direct consequence of having used advantageously the Householder transformation and is obtained at the expense of increased memory requirements and increased CPU time.This work was supported by Exxon Production Research Company, Houston, Texas. This paper is based partly on Refs. 1–4.     

Preconditioned iterative methods for a class of nonlinear eigenvalue problems

This paper proposes new iterative methods for the efficient computation of the smallest eigenvalue of symmetric nonlinear matrix eigenvalue problems of large order with a monotone dependence on the spectral parameter. Monotone nonlinear eigenvalue problems for differential equations have important applications in mechanics and physics. The discretization of these eigenvalue problems leads to nonlinear eigenvalue problems with very large sparse ill-conditioned matrices monotonically depending on the spectral parameter. To compute the smallest eigenvalue of large-scale matrix nonlinear eigenvalue problems, we suggest preconditioned iterative methods: preconditioned simple iteration method, preconditioned steepest descent method, and preconditioned conjugate gradient method. These methods use only matrix-vector multiplications, preconditioner-vector multiplications, linear operations with vectors, and inner products of vectors. We investigate the convergence and derive grid-independent error estimates for these methods. Numerical experiments demonstrate the practical effectiveness of the proposed methods for a model problem.     

Correlations for superpositions and decimations of Laguerre and Jacobi orthogonal matrix ensembles with a parameter

A superposition of a matrix ensemble refers to the ensemble constructed from two independent copies of the original, while a decimation refers to the formation of a new ensemble by observing only every second eigenvalue. In the cases of the classical matrix ensembles with orthogonal symmetry, it is known that forming superpositions and decimations gives rise to classical matrix ensembles with unitary and symplectic symmetry. The basic identities expressing these facts can be extended to include a parameter, which in turn provides us with probability density functions which we take as the definition of special parameter dependent matrix ensembles. The parameter dependent ensembles relating to superpositions interpolate between superimposed orthogonal ensembles and a unitary ensemble, while the parameter dependent ensembles relating to decimations interpolate between an orthogonal ensemble with an even number of eigenvalues and a symplectic ensemble of half the number of eigenvalues. By the construction of new families of biorthogonal and skew orthogonal polynomials, we are able to compute the corresponding correlation functions, both in the finite system and in various scaled limits. Specializing back to the cases of orthogonal and symplectic symmetry, we find that our results imply different functional forms to those known previously.