Explanation for PIDrobust methodology to a layman

[marellasunny]

Can someone explain what the PID robust methodology is all about in simple words?http://www.irt.rwth-aachen.de/en/fuer-studierende/downloads/pidrobust/

One of the advantages of the method stated is that "capability to process plants with delay time
".
Whats so great about that? Usual feedback loops can inculcate delay times in the system of differential equations somehow.I'm just a novice in control systems.

 

[maajdl]

Did they say there is something so great about that?

 

[marellasunny]

The PIDrobust Toolbox contains the following functionalities und features:
1.capability to process plants with delay time
2.capability to support plant uncertainties
3.adjusting the controller parameters based on maximum robustness
4.adjusting the controller parameters in a interactiv editable root locus plot
5.loop shaping based on Least Spuare Error (LSQ) Method
6.loop shaping based on H infinity Method
7.parameter optimization based on Ziegler-Nichols Step Response Method
8.parameter optimization based on CHR Response Method
9.parameter optimization based on Cohen-Conn Method
10.parameter optimization based on IMC Design Technique
11.analysis methods (step, impuls, and nyquist response)
12.plenty examples
13.ability to export all plots for further use

 

[marellasunny]

What's the significance of the 1st point above?

 

[maajdl]

I am not a specialist.

First, I suggest you to test a conventional PID to control a delay system, to learn what might happen.
Intuitively, this may lead to instabilities.
The corrections will be increased without use during the delay time.
Of course, the instability can always be reduced by reducing the feedback.
But reducing the feedback will also reduce the efficiency and in the end keep a "lag error".

Avoiding this instability while keeping efficiency needs a better way to calculate the correction.
The correction must take previous corrections into account, and specially their expected effect on the system output.
Intuitively, this suggest the need for a king of "memory" associated with the PID.
I think, it is more precisely a model a the "plant" that is needed to calculated the correction.
A correction would be needed if the real system shows a difference with the model system.

In computer-based feedback systems, algorithms might easily be adapted for delays.
In a high frequency range, for analog system, delay lines are available, and can also act a a kind of memory.
Such a delay line could be used to "model the system" too and make a delay-PID feedback possible.

However, when no delay lines are available, how can we proceed?

Without more information from the Aachen paper, it is hard for me to guess more about their system.
However, some suggestions might be found there:

http://msc.berkeley.edu/PID/modernPID3-delay.pdf

It might -in the end- be a matter of approximating the delay line in some way.
(no way to approximate an "advance line of course!)

See also some thesis on this topic, like:

http://scholarworks.sjsu.edu/cgi/viewcontent.cgi?article=5034&context=etd_theses

 

[skeptic2]

What's the significance of the 1st point above?

The significance is that if there is a significant delay between the application of the correcting signal and when that correction reaches the sensor, the system will oscillate. A PD controller can prevent that situation by adding the differential of the signal to the proportional signal which causes the controller to anticipate and compensate for the delay.

Imagine a tank filled with a liquid that must be heated and maintained at a certain temperature. If the temperature sensor is removed from the heating element there will be a delay between applying heat and the sensor sensing the correct temperature. In addition to preventing oscillation of the temperature, a PD controller is more robust in that as liquid is removed from the tank, the remaining liquid heats up and cools down more rapidly. That means that the differential part of the signal increases and in turn increases the compensation of the delay. Without it, a system that is stable at one volume of liquid would become unstable at another volume.