Parametric sensitivity analysis and real-time optimal control using TransWORHP
Abstract
Algorithms for the numerical solution of discretized optimal control problems are already used successfully in many areas of application, e.g. robotics, aerospace engineering, chemistry, or automotive industry. However, to improve the industrial applicability in case of perturbations, strategies for realtime solutions are necessary. For an optimal solution, the recent module WORHP Zen provides parametric sensitivity differentials. These values describe how the unperturbed optimal solution would change if the vector of parameter is modified slightly. We will discuss methods, how these off-line values can also be used for a fast on-line approximation of the perturbed solution. The cost-efficiency and robustness of these iterative methods will be illustrated by applied examples using tools available in TransWORHP.
Origin : Files produced by the author(s)