Control System Modelling Process

[DavidBaratheon]

Hi everyone, I am a fresh graduate working in industry (BEng Electronics Background).

I am trying to understand the control system modelling design process a bit better.

I am planning to use a LabVIEW programmed PID controller, but first want to model the system so that I have my PI values and can confirm I will get the correct behavior from the system.

What pieces of information do I need to gather in order to put together a document modelling the control system? I presume I will need some sort of block diagram showing the system, for example a motor output and some laser sensors for measuring movement going back into the closed loop control system.

If the motion created a linear motion (e.g. a lift going up and down) rather than a rotational one, and a sensor was able to tell the distance that the "lift" had moved, how would I go about modelling that system and selecting the right PID values for the control system?

Any insight into the design and modelling process will be greatly appreciated. I have some university level experience such as a quadcopter and IMU sensor but most of us struggled to get it working as it was an embedded systems module so not much focus on the actual control system modelling part. We had a control module but it was very basic. So am trying to get my skills up to scratch in industry. I have a copy of MATLAB if required. Any tutorials, examples, brief run through of the modelling process used in industry or other advice etc will be greatly appreciated.

Many thanks.

 

[papab]

I think the your question is too broad & difficult to answer. A controls text book would be a good place to start. Do you have Simulink? They have quite a bit of tutorials and examples that could help too.

 

[DavidBaratheon]

I do have Simulink. If you have any specific tutorials you could recommend, that would be great.

I do have some control systems text books, and completed a module on control systems as part of my electronics degree. But I felt I only really came away knowing the basics. Not well enough to work on an industry project.

How would you approach the lift analogy? Is there any way I can make the question a bit more specific?

 

[IRstuff]

Have you looked at

http://ctms.engin.umich.edu/CTMS/index.php?aux=Home

TTFN
I can do absolutely anything. I'm an expert!
homework forum: //

http://www.engineering.com/AskForum/aff/32.aspx

faq731-376 forum1529

 

[papab]

In many respects it doesn't matter if the motion is linear or rotational. Work thru some of the tutorials IRstuff pointed to.

 

[DavidBaratheon]

I am currently working through them. Many thanks.

Wouldn't the equation representing the system be different? In University we didn't learn how to model a mechanical system, so I am not sure how to actually represent the mechanical system as a model. Without an accurate model of the system, I would not be able to simulate the system and then try some PID values to see what gives me an optimum settling time and overshoot. The LabVIEW PID controller requires P, I and D values as inputs. So essentially knowing how to model a mechanical system is essential for being able to implement a control system.

 

[greg2835]

Yes the first thing you should do as a controls engineer is to determine an appropriate model for the system you are trying to control. When you get into higher level classes you will find that there are many ways to do this but for simple linear and rotational systems you can use the governing equations (i.e. Newtons's second law, DC motor equations, etc). Only when this model is determined can you go forth and calculate PID values (or PI values in your case).

I have found that in EE controls classes they do not appropriately cover how to model mechanical systems. But you are correct, before you take this to LabVIEW (or MATLAB) you have to know the model of the system you are trying to control.

I am happy to offer more assistance if needed.

Best!

 

[PNachtwey]

First you need to identify the system or plant you are trying to control.

I have found that in EE controls classes they do not appropriately cover how to model mechanical systems.

True but that should be the job of the mechanical engineer or designer. I have yet to see a mechanical system with a transfer function stamped on it.

I often need to model a mechanical/motor/hydraulic system. I don't have access to things like friction, mass, force etc. The only thing I can do is excite the system and model it in terms of gains, damping factors, natural frequency or time constant. System identification is something I had to learn on my own. I wasn't taught that in college.




Peter Nachtwey
Delta Computer Systems

http://www.deltamotion.com