2.0 KiB
History
Where did Robust Control come from?
After the beginnings of modern control and the development of optimal control, John Doyle released a paper in 1978 titled Guaranteed Margins for LQG regulators. This is a less than one page paper that basically gave birth to the robust control field, with a three word abstract: "There are none." I'm working out the kinks in this one (Feedback Control Theory), but essentially the gaussian part of the LQG is what destroys the guaranteed part of the phase and gain margins. The additional estimator involved can really screw with things.
I should add some context: 4 Qualifying Exam/3 Notes/Feedback Control Theory.
What does Robust Control do?
Robust control works with What is gain scheduling?
What are the limits?
Not particularly a limit but something to look at: there are a ton of papers that use the word 'robust' but aren't actually doing textbook robust control as Doyle puts it. Instead, they're doing some kind of formal methods and calling it robust because...? Who knows. Here's some examples: farzanRobustControlSynthesis2020
Limitation: Using automated design tools for robust control of SISO systems has its benefits outweighed by the labor involved in creating the weighting transfer functions [@atsumiModifiedBodePlots2012] atsumiModifiedBodePlots2012.
There is some work going on that tries to deal with this @atsumiModifiedBodePlots2012 . These people have made a tool that can create transfer functions using an upper bound of \Delta and W
Limitation: Disk unstructured uncertainty cannot lend itself to creating individual examples of perturbed plants. It is not as simple as picking a plant that is within the robust control disk, because the transfer function that actually gets you there is lost in the abstraction. Or perhaps generalized.
Limitation: Verifying the model of a controller is not sufficient to verify the implementation of one.
- Find sources that support this claim⏳ 2024-10-16