Nonlinear Approximate Optimal Control Using C/GMRES

In the resent years it became popular to solve control problems by using optimization. The most widespread form of such control strategies is model predictive control or MPC. The input to the plant is calculated in a way, that a cost function is minimized over a certain horizon of time. In this thesis a known method for solving an optimization problem considering continuous-time state space models is transformed into a method to solve an optimization problem, bounded by discrete-time descriptions of the system, described by a state space model or an input-output model. The developed optimization routines are applied to a rather simple toy example to demonstrate their functionality and to the airpath of a passenger car's diesel engine on a test bench.