Necessary conditions in optimal control without differentiability

被引:0
|
作者
Loewen, PD [1 ]
Vinter, RB [1 ]
Zheng, H [1 ]
机构
[1] Univ British Columbia, Dept Math, Vancouver, BC V6T 1Z2, Canada
关键词
D O I
暂无
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
In traditional proofs of the Maximum Principle, continuous differentiability, or at least Lipschitz continuity, of the dynamic constraint with respect to the state variable is invoked, along arbitrary control functions. Recently Sussmann, following up ideas proposed by Lojasiewicz, has demon strated the validity of the Maximum Principle in circumstances when the dynamic constraint is Lipschitz continuous merely along the optimal control function. We provide a simple derivation of the Maximum Principle, under this milder hypothesis, for problems with unilateral state constraints and where the right endpoint constraint takes the form of a family of functional inequalities. Eke land's Theorem is used to construct a sequence of perturbed 'Lipschitz' optimal control problems. The Maximum Principle (for non-Lipschitz data off the optimal control) is then proved by applying the standard Maximum Principle to each of the perturbed problems and passing to the limit.
引用
收藏
页码:19 / 24
页数:6
相关论文
共 50 条