Matlab code to determine the search domain constraints

Matlab code to determine the seach domain constraints for Bayesian optimisation. The constraint are selected to prevent unstable iterations during the on-line tuning process.

To expand the search space around the tuning parameters of the known controller Kα, a robust stability analysis is conducted on an initial set of constraints to determine how much uncertainty over and above the initial constraints can be tolerated. The initial gain constraints are cautiously selected as a factor of 2 in the direction of instability, and boldly selected as a factor of 0.2 in the opposite direction. Selection of the initial integral time constraints follows the opposite approach, i.e. a factor of 0.5 in the direction of instability and a factor of 5 in the opposite direction. The Robust Control Toolbox of MATLAB provides the stability margins for the uncertain system incorporating controllers with uncertain tuning parameters. The robust stability analysis provides the maximum parameter uncertainty that can be tolerated before the worst-case uncertainty yields instability.