Control Bootcamp: Sensitivity and Robustness

[@ft6637]

Excellent video as always. It just so happened that I found a paper of yours on closed loop turbulence control and saw that it was in cooperation with the TU Braunschweig (Germany). So I took the time and googled the co-author Bernd Noack for fun. And I cannot believe it, that he was born in Korbach, which is like 30km from where I live :D And studied in Göttingen not far from here either. So I am just again shocked, how small this world is. And even more in such special fields like control engineering. :D

[@boxiao6830]

Combine the Nuyquist Theory, does that mean there's a limit where we could place the poles when we use pole placement method?

[@MrBubblegumx]

How would one analyze stability in a system that changes between different phases but is linear in each one?

[@petersharman8013]

Thanks for an excellent video series. Can you recommend a good text book for classical control ?

[@DefinitelyNotJoao]

Since we are dealing with linear systems, by plugging in all frequencies i.e. the y axis why don't we also get all frequencies as outputs (with phase shifts)? Shouldn't the plot at 2:30 span the entire y axis? You talk about this in "Three Equivalent Representations of Linear Systems". Is it because the change in magnitude scales the points so that they become bounded? I hope you understand what I mean...

[@theyeah132]

The plot drawn at 2:30 is a Nyquist plot. The y-axis is not the frequency rather just the imaginary axis.

[@wangbo2276]

Thank you for really value video! I have several questions to ask you sincerely.The motivation of the robustness control is to deal with the uncertainty of the system which the LQG is hardly to solved. However, if the control contain no estimator, the simple optimal control with only LQR can also deal with the uncertainty of the system. In the absence of an estimator, is there any difference between optimal control and robust control in dealing with system uncertainty? If the estimator is considered, where should it be placed in robust control?

[@theyeah132]

As far as I know both LQR and LQG robustness can be checked by rewriting the block diagram and formulating the loop-transfer function. In the case of LQG the robustness has to be checked afterwards since there are no guarantees on the robustness of the system (see John Doyle's "Guaranteed margins for LQG Regulators").

[@evanparshall1323]

Why do you talk about there being peaks in S when the plot shows the peak being in T?

[@Dany_B_Good]

If you watched the next video of the series you saw that he fixed this error.