With specific forms of parametric uncertainty, I am getting incorrect results when analyzing the stability using robstab in the Robust Control Toolbox. The destabilizing perturbation returned does not result in an unstable system. Additionally, the identified bounds are incorrect.
It seems the issue is limited to situations in which the ureal parameter does not have equal +/- percentages. An example is shown below of a 1st-order low-pass filter with uncertain time constant. This system should be unstable for negative tau, and stable for positive tau.
% Example code of incorrect robstab results
tau = ureal('tau',1,'Range',[0.75 1.5]);disp(tau);G = tf(1,[tau 1]);[STABMARG, WCU, INFO] = robstab(G,opts);
The resulting upper bound is 300% (I think it should be 400%) and I'm told tau = 1 is destabilizing (which is not).
Best Answer