基于回归神经网络的非线性时变系统辨识

为克服基于前馈神经网络的非线性系统辨识算法存在需预先估计系统输入输出滞后阶数的缺陷，提出一种基于回归神经网络的非线性时变系统的辨识算法，针对现有的回归网络学习算法大多采用梯度算法，收敛速度缓慢问题，提出一种具有快速收敛性的扩展卡尔曼滤波学习算法，大大提高了学习收敛速度，并推导了一种基于单个神经元的局部化算法，减少了计算量，仿真实例证明，所提出的算法是有效的。     

基于Haar 小波变换的连续时间系统鲁棒参数辨识

给出用Ｈａａｒ小波对连续时间系统的鲁棒辨识方法。该方法在用Ｈａａｒ小波对系统输入和输入展开时,通过极小化一个鲁棒指标来减少噪声对展开系数的影响。因此对连续时间系统可获得鲁棒参数估计。仿真结果表明了该算法的有效性。     

Delay dependent stabilization of linear time‐varying system with time delay

This paper investigates both stability and stabilization dependent on the delay of a class of time‐varying linear systems with a constant point time delay. The matrices, describing the state space dynamics, are parameterized by time‐varying function matrices. A numerical example is given in order to verify the theoretical results. Copyright © 2009 John Wiley and Sons Asia Pte Ltd and Chinese Automatic Control Society     

Singularly perturbed implicit control law for linear time varying SISO systems

This paper considers the problem of stabilizing a single‐input single‐output linear time varying system using a low order controller and a reference model. The closed loop is a linear singularly perturbed system with uniform asymptotic stability behavior. We calculate bounds ? ∈ (0,? ^* ) as in Kokotovi?'s book, such that the uniform asymptotic stability of the singularly perturbed system is guaranteed. We show how to design a control law such that the system dynamics is assigned by a Hurwitz polynomial with constant coefficients. Copyright © 2013 John Wiley & Sons, Ltd.     

基于局部线性嵌入和Haar小波的人脸识别方法

为了抑制局部线性嵌入算法对噪音的敏感性,结合Haar小波变换,提出了一种人脸识别的新方法。利用Haar小波变换将原始图像数据分解为高频分量和低频分量,忽略水平高频与垂直高频分量,并将低频分量按行堆叠的方式引入其原始图像数据中。通过LLE对该图像数据进行降维,求得训练和测试样本各自对应的矩阵。依据最近邻准则完成人脸识别。基于ORL与Sheffield人脸数据库的实验结果表明了该方法对改善传统LLE算法识别率的有效性。     

基于多核计算平台和高速缓存感知的Haar小波变换算法

针对目前多核计算平台的普及性以及多核平台的强大计算能力,通过充分利用高速缓存的工作原理以及多线程程序设计的优势,提出了一种在多核平台上高速缓存优化的并行Haar小波计算算法.通过测定算法运行过程中高速缓存缺失率以及算法在不同多核计算平台上的运算性能,反映了该算法极大地降低了缺失率和减少了计算时间.如在数据规模8192条件下缺失率从95%降低到8.37%,计算时间从4.35s减至0.89s.由此证明了该计算方法具有高速且可移植的特性.     

采用SIMT结构GPU的二维离散哈尔小波变换的优化

为了解决CPU环境下小波变换在运行时对高分辨率图片处理速度较慢的问题,利用GPU有大量可编程核心的特点,针对二维离散哈尔小波变换进行了在SIMT(单指令多线程)体系结构GPU环境下的并行推导,同时调整GPU的逻辑布局,将数据分割,更改了数据同步方式,并且采用了虚拟寻址,将速度进一步提升到了0.92ms,比CPU环境下效率提升51.1%,比SIMD架构效率提升16.3%,效果显著,满足实时性要求。     

Regularizations of linear time varying singular systems

Singular systems of differential equations arising in reduced-order models are often regularized by the introduction of small parameters. The types of regularization used for linear time invariant systems need not work on linear time varying systems. This paper gives examples and discusses this behavior. Conditions under which the standard type of regularization works are developed.     

Solution of variational problems via Haar orthonormal wavelet direct method

This article establishes a clear procedure for the variational problem solution via the Haar orthonormal wavelet technique. The variational problems are solved by means of the direct method using the Haar orthonormal wavelets and reduced to the solution of algebraic equations. The local property of the Haar orthonormal wavelets is fully applied to shorten the calculation process in the task. Three illustrative examples and a practical application to a heat conduction problem are included.     

Delay‐dependent conditions for finite time stability of linear time‐varying systems with delay

This paper investigates the problem of finite time stability of linear time‐varying system with delay. By constructing an augmented time‐varying Lyapunov functional and using the Wirtinger‐type inequality deductively, delay‐dependent finite time stability conditions are derived and presented in terms of differential linear matrix inequalities (DLMIs). Then, the DLMIs are transformed into a series of recursive linear matrix inequalities (RLMIs) by discretizing the time interval into equally spaced time distances, and an algorithm is given to solve the RLMIs. Examples illustrate the feasibility and effectiveness of the proposed method.     

Numerical solution of linear Fredholm integral equations system by rationalized Haar functions method

In this paper rationalized Haar functions are developed to approximate the solutions of the linear Fredholm integral equations system. Properties of rationalized Haar functions are first presented, the operational matrix of the product of rationalized Haar functions vector is utilized to reduce the computation of Fredholm integral equations system to some algebraic equations. Finally, numerical result are given which support the theoretical results.     

On new estimates for Lyapunov exponents of discrete time varying linear systems

In this paper, we propose certain new bounds for the Lyapunov exponents of discrete time varying linear systems. The bounds are expressed in terms of spectral radii of matrix coefficients and therefore may be used to establish the exponential stability of time varying system on the basis of eigenvalues of individual coefficient. This approach is known in the literature as frozen time method.     

Haar小波滤波器序列只有两个非零项的论证

Haar小波是最简单的紧支集正交小波（Daubechies小波）,其滤波器序列较短,在图像处理等诸多领域都有广泛的应用。由Daubechies小波的构造理论可知,现有的正交小波是在比较特殊的前提下得到的,则Haar小波的滤波器系数序列的唯一确定性受到质疑。以多分辨分析为基础,在时域对Haar小波滤波器系数序列的唯一性进行了论证,即证明了Haar小波滤波器序列只有两个非零项,这对促进小波的理论完善与应用研究具有十分重要的意义。     

基于Haar小波变换和分块DCT的人脸识别

提出一种新的基于二维离散小波分解和分块离散余弦变换的降维方法.该方法与模式识别领域中用于特征提取和降维的PCA-LDA方法进行了比较.结果表明,此方法与PCA-LDA方法在识别率上大体相当,但它比其更具有计算量小、降维速度快的优点.因此,该方法对于人脸识别是一种有效降维手段.     

Identification of linear systems using polynomial kernels in the frequency domain

In prior work we presented an identification algorithm using polynomials in the time domain. In this article, we extend this algorithm to include polynomials in the frequency domain. A polynomial is used to represent the imaginary part of the Fourier transform of the impulse response. The Hilbert transform relationship is used to compute the real part of the frequency response and hence the complete process model. The polynomial parameters are computed based on the computationally efficient linear least square method. The order of the polynomial is estimated based on residue decrement. Simulated and experimental results show the effectiveness of this method, particularly for short input/output data sequence with high signal to noise ratio. The frequency domain polynomial model complements the time domain methods since it can provide a good estimate of the time to steady state for time domain FIR (finite impulse response) models. Confidence limits in time or frequency domain can be computed using this approach. Noise rejection properties of the algorithm are illustrated using data from both simulated and real processes.     

一种自动调整遗忘因子的快速时变参数辨识方法

本文介绍了一种自动调整遗忘因子的快速时变参数辨识方法.这种方法主要是在文献[1,2]的基础上改进了遗忘因子的选取方法,引入了自动调整遗忘因子,以适应导弹控制系统等快速时变系统的参数估计要求.数字仿真结果表明,这种辨识方法对于诸如参数的指数变化、斜坡变化、阶跃变化、正弦变化、方波变化以及由这些变化形式构成的混合变化,都有比较好的辨识效果.     

A technique for dynamic battery model identification in automotive applications using linear parameter varying structures

In this paper, a rapid calibration procedure for identifying the parameters of a dynamic model of batteries for use in automotive applications is described. The dynamic model is a phenomenological model based on an equivalent circuit model with varying parameters that are linear spline functions of the state of charge (SoC). The model identification process is done in a layered fashion: a two step optimization process using a genetic algorithm (GA) is used to optimize the parameters of the model over an experimental data set that encompasses the operating conditions of interest for the batteries. The level of accuracy obtained with this procedure is comparable to other black/gray box techniques, while requiring very little calibration effort. The process has been applied to both lithium ion and NiMH chemistries with good results. An extension of this technique to identify a model with both SoC and temperature dependence is discussed.     

A computational method for solving optimal control and parameter estimation of linear systems using Haar wavelets

In this article, a computational method based on Haar wavelet in time-domain for solving the problem of optimal control of the linear time invariant systems for any finite time interval is proposed. Haar wavelet integral operational matrix and the properties of Kronecker product are utilized to find the approximated optimal trajectory and optimal control law of the linear systems with respect to a quadratic cost function by solving only the linear algebraic equations. It is shown that parameter estimation of linear system can be done easily using the idea proposed. On the basis of Haar function properties, the results of the article, which include the time information, are illustrated in two examples.     

Robustness analysis of linear periodic time-varying systems subject to structured uncertainty

In this paper, we show how Floquet theory may be combined with a technique known as Lifting to cast a linear periodically time-varying system subject to structured linear time invariant uncertainty in the form of a linear fractional transformation. The stability and performance robustness of the resulting system may then be analysed using standard μ-analysis methods. A significant advantage of the proposed approach is that it allows the computation of a worst-case destabilising uncertainty combination which may be used to estimate the conservatism of the computed robustness margin. An example is given to illustrate the application of the proposed approach.