C06: Port-Hamiltonian methods for optimal control of large-scale systems
Principal investigators:
Prof. Dr. Hannes Gernandt; Prof. Dr. Manuel Schaller
Project description:
We investigate the previously unnoticed port-Hamiltonian structure of iterative numerical solvers of large optimal control problems. Based on this, optimization-based controllers are formulated in which the iterative solver of the optimal control problem is coupled with a system to be controlled. The decisive advantage of this new approach is that the inherent dissipativity of port-Hamiltonian systems guarantees the stability of the closed control loop, especially if, for example, learning-based surrogate models are used in the optimal control problem.