Journal of Convex Analysis 23 (2016), No. 1, 263--290
Copyright Heldermann Verlag 2016
About Moreau-Yosida Regularization of the Minimal Time Crisis Problem
Terence Bayen
Institut Mathématiques, Université Montpellier, CC 051, 34095 Montpellier Cedex 5, France
tbayen@math.univ-montp2.fr
Alain Rapaport
Montpellier SupAgro, 2 Place Pierre Viala, 34060 Montpellier Cedex 2, France
rapaport@montpellier.inra.fr
We study an optimal control problem where the cost functional to be minimized represents the so-called time of crisis, i.e. the time spent by a trajectory solution of a control system outside a given set K. This functional can be expressed using the characteristic function of K that is discontinuous preventing the use of the standard Maximum Principle. We consider a regularization scheme of the problem based on the Moreau-Yosida approximation of the indicator function of K. We prove the convergence of an optimal sequence for the approximated problem to an optimal solution of the original problem. We then investigate the convergence of the adjoint vector given by Pontryagin's Principle when the regularization parameter goes to zero. Finally, we provide an example illustrating the convergence property and we compute explicitly an optimal feedback policy and the value function.
Keywords: Optimal control, Pontryagin Maximum Principle, Hybrid Maximum Principle, Regularization.
[ Fulltext-pdf (281 KB)] for subscribers only.