A local search scheme for the inequality-constrained optimal control problem (Q2117613): Difference between revisions

In this paper, the author reduces the original optimal control (OC) problem to the problem without inequality constraints via the Exact Penalization Theory. The penalized problem is linearized with respect to the basic nonconvexity. The nonconvex optimal control problem with the cost functional and inequality constraint given by the functionals of Bolza is addressed. All the functions in the statement of the problem are state-DC. The control system is state-linear. With the help of the Exact Penalization Theory, the state-DC form of the penalized cost functional is proposed and, using the linearization at every current iteration with respect to the basic nonconvexity of the penalized problem, the linearized optimal control problem is studied. Since special local search procedures play a very important and useful role in the global search in nonconvex optimization problems, providing not only the KKT points or the Pontryaguin's extremal, but often more strong control processes. On this basis, a general scheme of the special Local Search Method with a varying penalty parameter is developed. Further, the convergence of the proposed scheme is studied. For the entire collection see [Zbl 1482.90002].