-
在本文中,设控制系统为
性能指标为
设
$\mathit{\boldsymbol{A}}( \bullet ) \in {L^\infty }\left( {[0, T];{\mathbb{R}^{n \times n}}}\right), $ $\mathit{\boldsymbol{B}}( \bullet ) \in {L^\infty }\left( {[0, T];{\mathbb{R}^{n \times m}}} \right),\mathit{\boldsymbol{R}}( \bullet ) \in {L^\infty }\left( {[0, T];{S^n}} \right), \mathit{\boldsymbol{M}}( \bullet ) \in {L^\infty }([0, T];{\mathbb{R}^{m \times n}}), $ $\mathit{\boldsymbol{N}}( \bullet ) \in {L^\infty }\left( {[0, T];{S^m}} \right), \mathit{\boldsymbol{G}} \in S_ + ^n$ 并且$\boldsymbol{R}(\bullet), \boldsymbol{M}(\bullet), \boldsymbol{N}(\bullet)$ 满足其中:Sn表示n×n阶对称矩阵全体,S+n表示n×n阶半正定对称矩阵全体.
设控制集U为
$\mathbb{R}^{m}$ 中的有界闭凸集,可行控制集$\mathscr{U}$ ad定义为由以上假设知
$J(\boldsymbol{u}(\bullet))$ 为凸函数.本文考虑的约束线性二次最优控制问题(CLQ)为:求u(·)∈
$\mathscr{U}_{a d}$ ,使得满足(2)式的u(·)和(1)式关于u(·)的解x(·)组成的(x(·),u(·))称为最优对.
由Pontryagin最大值原理[1],最优对(x(·,u(·))若存在,必满足一阶最优性条件:
其中
称为伴随方程.
在L2上定义范数‖·‖2和内积〈φ,ψ〉L2,
设
$\mathit{\boldsymbol{x}} \circ \mathit{\boldsymbol{u}}( \bullet )$ 为(1)式中微分方程关于u(·)的解,$\mathit{\boldsymbol{p}} \circ \mathit{\boldsymbol{u}}( \bullet )$ 为(4)式中伴随方程关于u(·)和$\mathit{\boldsymbol{x}} \circ \mathit{\boldsymbol{u}}( \bullet )$ 的解.定义映射$F\circ \bullet : \mathscr{U}_{a d} \longrightarrow L^{2}\left(0, T ; \mathbb{R}^{m}\right)$ 于是(3)式可转化为抽象变分不等式VI(F,
$\mathscr{U}_{a d}$ ):求解u∈$\mathscr{U}_{a d}$ 使得这样,一阶必要性条件可等价写成变分不等式VI(F,
$\mathscr{U}_{a d}$ )的形式,而在凸性的条件下,一阶必要条件为充要条件(定理2),故可以将最优控制问题等价转化为变分不等式问题.进而利用变分不等式问题的Tikhonov正则化方法来证明CLQ问题的正则性.
The Tikhonov Regualization of the Optimal Control Problem
-
摘要: 主要研究了约束线性二次最优控制问题.通过一阶最优性条件将它等价地转化为单调变分不等式问题,并利用变分不等式的Tikhonov正则化方法研究了约束线性二次最优控制问题的正则化,证明了扰动问题的解收敛到原问题的最小范数解.
-
关键词:
- CLQ问题 /
- 变分不等式 /
- Tikhonov正则化 /
- 扰动问题
Abstract: In this paper, the problem of linear quadratic (LQ) optimal control with constraint is discussed. First, this problem is equivalently converted to a monotonic variational inequality problem through first-order optimality conditions. Then, by the Tikhonov regularization method of variational inequalities, the regularization of the problem is studied. Finally, we prove that the solution of the perturbation problem converges to the minimum norm solution of the original problem.-
Key words:
- CLQ problem /
- variational inequality /
- Tikhonov regularization /
- perturbation problem .
-
[1] CHAMBERS M L, PONTRYAGIN L S, BOLTYANSKⅡ V G, et al. The Mathematical Theory of Optimal Processes[M]. Oxford:Pergamon Press, 1964. [2] FACCHINEI F, PANG J S. Finite-Dimensional Variational Inequalities and Complementarity Problems[M]. New York:Springer, 2004. [3] 张亮, 吴至友.非凸变分不等式的四步投影算法及其收敛性分析[J].西南大学学报(自然科学版), 2014, 36(10):109-113. doi: http://xbgjxt.swu.edu.cn/jsuns/jsuns/ch/reader/view_abstract.aspx?file_no=2014-10-109&flag=1 [4] 许微, 彭建文.基于投影交替方向法求解结构型单调变分不等式[J].西南大学学报(自然科学版), 2016, 38(1):90-97. doi: http://xbgjxt.swu.edu.cn/jsuns/jsuns/ch/reader/view_abstract.aspx?file_no=201601014&flag=1 [5] MINTY G J. Monotone (Nonlinear) Operators in Hilbert Space[J]. Duke Mathematical Journal, 1962, 29(3):341-346. doi: 10.1215/S0012-7094-62-02933-2 [6] BROWDER F E. Existence and Approximation of Solutions of Nonlinear Variational Inequalities[J]. Proceedings of the National Academy of Sciences, 1966, 56(4):1080-1086. doi: 10.1073/pnas.56.4.1080 [7] doi: http://d.old.wanfangdata.com.cn/OAPaper/oai_doaj-articles_7806eac68c3a201385a43c8472d12b87 HE Y R. The Tikhonov Regularization Method for Set-Valued Variational Inequalities[J]. Abstract and Applied Analysis, 2012, 2012:1-10. [8] QI H D. Tikhonov Regularization Methods for Variational Inequality Problems[J]. Journal of Optimization Theory and Applications, 1999, 102(1):193-201. doi: 10.1023/A:1021802830910 [9] LUO X P. Tikhonov Regularization Methods for Inverse Variational Inequalities[J]. Optimization Letters, 2014, 8(3):877-887. doi: 10.1007/s11590-013-0643-4
计量
- 文章访问数: 823
- HTML全文浏览数: 655
- PDF下载数: 141
- 施引文献: 0