Optimalitätsbedingungen, parametrische Sensitivitätsanalyse und Echtzeitanpassung optimaler Regel- und Schätzverfahren

Abstract

This dissertation covers the derivation of necessary and sufficient optimality condition for linear quadratic regulator (LQR) problems. Based on this, parametric sensitivity methods and real time adaption techniques from nonlinear optimization theory are applied to LQR problems. In addition to that, through dual theory, these methods are extended to linear quadratic estimator problems as well as combined regulator-estimator problems. All derived adaption techniques are numerically tested for two examples.