Optimizing state trajectories using surrogate models with application on a mechatronic example

The classic design- and simulation methodologies, that are constituting today’s engineer main tools, fall behind with industry’s ever increasing complexity. The strive for technological advancement heralds new performance requirements and optimality remains no longer a concern limited to regime operation. Since the corresponding dynamic optimization problems incorporate accurate system models, the current techniques are plagued by the high computational weight these multi-disciplinary and highly dimensional system models bear with them. This imbalance advocates for the need to adapt the existing approaches. In this study we propose an algorithmic framework as an extension of the direct transcription method, which has already proven its usefulness concerning this matter. It is suggested to construct a surrogate model of the derivative function that is iteratively refined in a region of interest. Thereafter the method will be illustrated on an academic yet nonlinear example