Note: Hespe, Christian and Werner, Herbert

Abstract: The aim of this project work is the implementation of a set of benchmark scenarios for nonlinear model predictive control (NMPC). The main focus is put on an algorithm that implements NMPC for plants that can be described as quasi-linear parameter-varying systems. In each benchmark scenario it is evaluated how this algorithm compares to state-of-the-art NMPC toolkits. Improvements to the algorithm are made by reconsidering its discretization method and presenting an alternative formulation for constant parts in ordinary differential equations. Benchmark results show that the algorithm computes a control input that is similar to its competitors in terms of optimality. The algorithm only shows superior in terms of computational performance when applied to small and unconstrained optimal control problems (OCP). Finally, an analysis of the algorithm's scalability shows its vulnerability regarding larger OCPs.