几类线性矩阵方程的矩阵级数解及其数值算法-运筹学与控制论专业论文.docxVIP

I I I II 摘 要 矩阵理论在统计学、梯形网络、运输理论、动态规划、控制理论和统计过滤 等领域中有着广泛的应用. 连续线性系统稳定分析和最优控制问题设计中的许 多问题常常可转化为线性矩阵方程的求解问题. 本文给出了 Sylvester 矩阵方程, 离散和连续 Lyapunov 矩阵方程等线性矩 阵方程的矩阵级数解, 并给出相应的数值算法. 本文分为三章: 第一章, 介绍了这几类线性矩阵方程的应用背景, 给出本文所用的记号和定 义. 第二章, 本章利用矩阵级数收敛的性质给出了线性矩阵方程 P = AP B + C 的矩阵级数解, 进一步根据离散 Lyapunov 矩阵方程 P = AP AT + Q 的系数矩阵 A 的特征值的模的不同范围, 给出了它的几类矩阵级数解. 并相应的设计了新的 求离散 Lyapunov 矩阵方程解的数值算法, 最后通过例子说明新算法的优越性. 第三章, 根据第二章的结论, 我们引入适当的参数, 将 Sylvester 线性矩阵方 程和连续 Lyapunov 矩阵方程等价变形为第二章中相应的线性矩阵方程, 并分别 给出了它们的矩阵级数解. 进一步设计了求解的数值算法. 数值例子验证了其优 越性. 关键词: 矩阵方程; 矩阵级数解; 参数; 数值算法 ABSTRACT The matrix theory is widely used in the areas of statistics, ladder networks, transport theory, dynamic programming, control theory and stochastic filtering. Several problems in the stability analysis of continuous time linear systems and in the design of optimal control systems are usually transformed to solve the linear matrix equations. In this paper, we present the matrix series solutions for some linear matrix equations, such as equation P = AP B + C, discrete and continuous Lyapunov matrix equations, and give some corresponding numeric algorithms. This paper mainly contains three chapters: In chapter one, we present some background knowledge of these linear ma- trix equations, and introduce some symbols and definition used in this paper. In chapter two, we obtain the matrix series solution for the linear matrix equation P = AP B + C, by using the property of the matrix series convergence. We present several matrix series solutions to discrete Lyapunov matrix equation P = AP AT + Q, while the module of the eigenvalues of the coefficient matrix A belong to different ranges. And design the corresponding numeric algorithms to the discrete Lyapunov matrix equation. Finally the superiority of the new numeric algorithms will be showed by numerical examples. In chapter three, based on the results of chapter two, and taking in a pa- rameter, we can change Sylvester matrix equation and the continuous Lyapunov matrix equatio