Control law synthesis for discrete time and finite horizon dynamical systems under set bounded uncertainty: formulation, controller duality, solvers

In this work a control law synthesis problem for discrete-time and finite control horizon systems under bounded model of uncertainty is considered. For discrete-time linear systems with uncertain parameters and classical information pattern the functional minimization problem is transformed to the parametric one by using theorem of iterated integral and notion of information sets. A criterion for lack of the dual effect is presented. For discrete-time nonlinear systems under nonclassical information pattern an information set approach is combined with parametrization of the control law by weighting coefficients of two layer neural network. Using this approach numerical solutions of example problem is obtained and detailed discussion of dual effect is presented