A semi‐smooth Newton method for constrained linear‐quadratic control problems
- 24 March 2003
- journal article
- research article
- Published by Wiley in ZAMM - Journal of Applied Mathematics and Mechanics / Zeitschrift für Angewandte Mathematik und Mechanik
- Vol. 83 (4) , 219-237
- https://doi.org/10.1002/zamm.200310030
Abstract
Es wird ein verallgemeinertes Newton‐Verfahren für die numerische Lösung von linear‐quadratischen Optimalsteuerprozessen mit zweiseitigen Restriktionen an die Steuerung eingeführt und analysiert. Bedingungen werden hergeleitet, unter denen der Algorithmus global konvergiert. Unter Verwendung von Glättungseigenschaften des Lösungsoperators der Zustandsgleichung und einem verallgemeinerten Ableitungsbegriff lassen sich lokal superlineare Konvergenzeigenschaften nachweisen. Umfassende numerische Tests unterstützen die theoretischen Ergebnisse, beweisen die Effizienz des neuen Verfahrens und liefern einen numerischen Nachweis für die Gitterunabhängigkeit des Algorithmus.A semi‐smooth Newton method for the numerical solution of linear‐quadratic optimal control problems with bilateral control constraints is introduced and analyzed. Conditional global convergence results are proved. Based on the smoothing property of the control‐to‐state operator and on generalized derivatives locally superlinear convergence is established. A report on comprehensive numerical tests supports the theoretical results, proves the efficiency of the new algorithm, and allows to argue mesh‐independence.Keywords
This publication has 15 references indexed in Scilit:
- Augmented Lagrangian methods for nonsmooth, convex optimization in Hilbert spacesNonlinear Analysis, 2000
- An SQP method for the optimal control of large-scale dynamical systemsJournal of Computational and Applied Mathematics, 2000
- SQP-methods for solving optimal control problems with control and state constraints: adjoint variables, sensitivity analysis and real-time controlJournal of Computational and Applied Mathematics, 2000
- A Comparison of a Moreau--Yosida-Based Active Set Strategy and Interior Point Methods for Constrained Optimal Control ProblemsSIAM Journal on Optimization, 2000
- Smoothing Methods and Semismooth Methods for Nondifferentiable Operator EquationsSIAM Journal on Numerical Analysis, 2000
- Second-Order Runge--Kutta Approximations in Control Constrained Optimal ControlSIAM Journal on Numerical Analysis, 2000
- Primal-Dual Strategy for Constrained Optimal Control ProblemsSIAM Journal on Control and Optimization, 1999
- Convergence of Lagrange-Newton method for control-state and pure state constrained optimal control problemsPublished by Springer Nature ,1996
- A nonsmooth version of Newton's methodMathematical Programming, 1993
- Stability and sensitivity of solutions to optimal control problems for systems with control appearing linearlyApplied Mathematics & Optimization, 1987