摘要
针对在时变系统中提出的广义Riccati矩阵方程约束解问题,基于共轭梯度算法原理建立了两种求广义Riccati矩阵方程异类约束解(对称和反对称解)的算法,即非精确牛顿修正共轭梯度算法(In-Newton-MCG算法)和非精确牛顿正交投影算法(In-Newton-OPA算法),并给出了两种算法收敛性结论和两种算法的数值实验.算例表明,In-Newton-MCG算法在一定条件下比In-Newton-OPA算法具有更高的计算效率.
Two algorithms called inexact Newton modified conjugate gradient algorithm(In-Newton-MCG algorithm)and inexact Newton orthogonal projection algorithm(In-Newton-OPA algorithm)are developed in this paper,the two algorithms are based on the principle of conjugate gradient algorithm,which are developed for solving the constrained solutions of generalized Riccati matrix equation in time-varying systems.The convergence results and numerical experiments of the two algorithms are given.Numerical experiments shows that the In-Newton-MCG algorithm is more efficient than the In-Newton-OPA algorithm under certain conditions.
作者
陈世军
CHEN Shijun(Yango University,Fuzhou 350015,China)
出处
《延边大学学报（自然科学版）》
CAS
2021年第2期120-125,130,共7页
Journal of Yanbian University（Natural Science Edition）
基金
福建省教育厅中青年教师教育科研项目(JAT190410)。
