Understanding Disk Margin | Robust Control, Part 2

[@drnsksai]

One of the excellent presentations in control

[@sirdonie1]

Thanks for this video, Brain! I was looking forward to watching more videos about robust control. I heard some stuff about robust control in 2012, when I was trying to do a final year project which was based on a previous robust control project. However, due to several factors, including available time, I couldn't delve into robust control theory. Now it's a great time for me to come back and start studying that again and, in the near future, implement anything driven by robust control. I hope the next video come soon :)

[@Kantha1980]

Excellent presentation. Glad and lucky to watch this artistic explanation of making common man understand robust controls. You deserve more Brain!

[@prasadelumalai946]

Ultimately its a juggle between system analysis methods and control design strategies and best control design is figuring out which analysis/design methods to use to control a given system .. Woof..🥵🥴.. The reality we wish to control is 💥

[@carlosrodriguez9480]

Brian, make a new series of videos on how to make interactive simulations. It would be SO HELPFUL!! The real power of Matlab is to help you explore a problem, a system, a design, but sometimes it´s a bit time-consuming to figure out how.

[@alex.ander.bmblbn]

one way i could think of doing that is to make up an N by N matrix of real and imaginary conponents and get a matrix of 0 and 1 for unstable and stable results and then meshing or surfing them with no edges but markers at all nodes. they will differ in colour, so you then take the xy view of it and here you go

[@abdosk4939]

Very nice explanation. Thanks

[@chintugurbani]

excellent way to visualise...

[@amaarquadri]

How did you generate the plot at 12:50? I can't find a MATLAB command that makes anything like it, and it seems like a very useful visualization tool!

[@Kantha1980]

It is made by interactively running a matlab script that handles mouse pointer location and corresponding value of the graph.

[@neerajkrishnang3916]

Did the specs for the diskmargin() function change? Can't seem to execute the commands shown in the video on Matlab 2020b.

[@alexanderskusnov5119]

What about delay? For example, in ventilation we have hot water valve in one place but air temperature sensor far from it.

[@Giammy12]

I might be wrong but delays produce phase shift, so that's what you need phase margin for

[@alexanderskusnov5119]

@@Giammy12 but this shift is not constant, it depends on frequency.

[@Giammy12]

@@alexanderskusnov5119 Basically a delay is equivalent to multplying the system by e^-(td*s), where td is the time delay. This affects the bode plot of the system: the magnitude stays the same but the phase decreases as frequency increases

[@Giammy12]

So you can calculate again the phase/disk margin by looking at the new value of the phase that corresponds to the critical frequency of the system

[@thiagor2656]

I think there is a video on the Matlab channel that adresses delay.

[@mehmetkilic9518]

Can you explain the "conservative" term, in brief, to understand more clearly, please?

[@mehmetkilic9518]

For both parametric and unstructured uncertainty, is it an important issue?

[@franciscos.2301]

​@@mehmetkilic9518 I don't think conservative has a specific meaning here. A conservative solution is a solution that errs on the side of safety or what is normal. It's a "less risky" solution

[@アセマ]

Why it is specifically a disk, couldn’t it be a square or any other form ?

[@Kantha1980]

disc chosen so that it can have a radius to account uniform variation around F=1

[@アセマ]

@@Kantha1980 thank you! After re-reading my question I realise that it was dumb lol :p

[@oldcowbb]

why fit a disk instead of finding the entire stable region?

[@Z0Gab]

Exactly what i thought of, why not a full elipse?

[@alex.ander.bmblbn]

@@Z0Gab why would you be looking for an elipse if its a triangle or something else?

[@Z0Gab]

@@alex.ander.bmblbn yeah my bad i didnt knew then, i think the problem of finding the entire stable region is that is not easily solvable neither interpretable

[@olivergriffiths1996]

Is there a name for the plot that shows the full complex stable and unstable regions ? The red and green plot? Looks like Brian made it manually by looping

[@wilurbean]

Is this Prof Seiler in the video? It sounds a lot like him

[@BrianBDouglas]

Peter Seiler isn't in the video, it's my voice you hear, however this video is based on the paper that Peter wrote along with Andrew Packard, and Pascal Gahinet.

[@wilurbean]

​@@BrianBDouglas I did see that at the end where it references your channel (which I have watched before). Its still uncanny how much you all sound similar on video. I'm taking Prof Seiler's senior level control systems class right now and he uses an inverted class style. That means we watch pre-recorded lecture videos ahead of class and work problems in class. I don't think I could distinguish the audio from either video series unless they were side by side.