How get PID parameters?

[PiramydHead]

Hello

I have hammer mill for grain milling, with manual dosator valve (feeder). Now i need an automatic control system for this mill.
I think i can build this system with simple PID controller but... How determine kp, ki, kd parameters without many experiments?
I think i need:
a) get system response function ( where x-time (secondes), y-hammer mill load (ampere) ) by opening manual dosator valve
for 70%(?) of full load.
b) calculate kp, ki, kd from function curve.
c) add pid controller (connect to dasator valve motor and to hammer mill motor current transformer) with calculated kp, ki, kd

Question: i'm in right way?

 

[Skogsgurra]

I think that you started in the wrong end. The controller parameters are the least problem. Find out what hardware you need first.

I cannot imagine that your mill needs much more than a P controller. What are your needs for precision? And how do you measure it?

Gunnar Englund

http://www.gke.org

 

[Ashereng]

There are software programs that also help determine PID parameters, but they also requrire system response data.

Here is a very rough guide that I use when I have to tune a control loop. Keep in mind that my systems are usually crude oil.

1) Set gain to 0.8
2) Set integral to 15 seconds per cycle (or 4 cycles per minute)
3) Let her rip.
4) If initial response time is too slow, increase gain by 0.2.
5) If accuracy is too far off, decrease integral by 5 seconds.
6) Let her rip.

If you have dead time, then you may need to play around with the derivative term. In 99.999% of my applications, PI is all I need.



This will get you close. If you want better, it takes more effort, as you've said.

 

[PNachtwey]

"There are software programs that also help determine PID parameters, but they also requrire system response data."
Pyramidhead, you should realize that Ashereng's suggest tuning parameters work on his particular oil system. Your system may be much different.
Pyramidhead, skogsgurra may be right about needing only a simple controller. There are very small PLCs that have PIDs that may be able to do your job. More info is needed.
I like PLCs that have a trend capability like a micrologix. This allows one to get the data necessary for tuning easily.

 

[PiramydHead]

Hmm. Hardware: hammer mill current i get from current transformers with transducer (4-20 mA. output).
Harder with feeder. I think we have to mount reverse gear motor. This motor have been controled by our controler ("digital output").
About controller. I want choose some like "West 6100+".
Thank you for new ides. However now i see that i have read more about feedback control principles

 

[powerjunx]

PirameadHead,
Visit EXPERTUNE.COM.. this might help you.

Check it out!

 

[itsmoked]

More specifically:

http://www.expertune.com/forum/viewforum.php?f=8

You probably won't want to afford their prices $$$$$$.

And I agree with skogs probably P only.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[powerjunx]

yeah, itsmoked.. youre right! PRICE is the main denuminator, it's pretty bit higher!

 

[PNachtwey]

If you want a magic bulllet then you must be prepared to pay for the knowlege. The PID gains can be calculated if plant gains and time constant(s) are known. One way you pay, the other way you know more.

The general procedure is to:
1 capture PV vs CV data in a file. This should show how the plant changes with respect to changes in SP.
2. calculate a model of the system. Google search for system identification. System identication will provide the plant gains and time constants ( poles and zeros ).
3. Calculate the controller gains as function of the plant gains poles and zeros and the desired response. This can be done by techniques such as pole placement or internal model control control direct synthesis.

A simple heat exchange example is on

http://www.controlguru.com