Author Topic: Design of PID parameters  (Read 3351 times)

0 Members and 1 Guest are viewing this topic.

Offline darkarticTopic starter

  • Beginner
  • *
  • Posts: 3
  • Helpful? 0
Design of PID parameters
« on: November 06, 2011, 11:29:00 AM »
Hi everyone,

I'm designing a PID controller for a 3 DOF robot arm, but first i need to control each link separately and then all of them together.

I just find information about tuning PID parameters but nothing about how to first theorically get them and then i can tune them

I was wondering which are the best design parameters for controlling a robot arm (overshoot, rise time, settling time) and THEIR VALUES, if what a i want is a balanced robot arm, fast and accurate enough.

I'll appreciate any opinions or if you can tell me where to find this information, remember i need the values of this parameters so i can start my design.

Greetings

« Last Edit: November 06, 2011, 11:35:44 AM by darkartic »
Logged

Offline rbtying

  • Supreme Robot
  • *****
  • Posts: 452
  • Helpful? 31
Re: Design of PID parameters
« Reply #1 on: November 07, 2011, 12:44:00 AM »
The reason why you won't find values (and why we won't give them) is because the parameters for PID can only be determined by perfectly modeling the system, which is especially difficult if you don't actually have it, and generally impractical even if you do. Most of the time, you use empirical methods to approximate the parameters.
Logged

Offline newInRobotics

  • Supreme Robot
  • *****
  • Posts: 1,015
  • Helpful? 48
  • N.I.R.
Re: Design of PID parameters
« Reply #2 on: November 07, 2011, 01:29:49 AM »
I found this PID Theory page very useful.
« Last Edit: November 17, 2011, 03:15:02 AM by newInRobotics »
Logged
"Computer games don't affect kids; I mean if Pac-Man affected us as kids, we'd all be running around in darkened rooms, munching magic pills and listening to repetitive electronic music." - Kristian W

Offline nm1

  • Jr. Member
  • **
  • Posts: 9
  • Helpful? 0
Re: Design of PID parameters
« Reply #3 on: November 16, 2011, 08:21:44 PM »
the Ziegler–Nichols method is commonly used if you don't know or want to model the system.
http://en.wikipedia.org/wiki/Ziegler%E2%80%93Nichols_method
Its quite simple too,
firstly you set ki and kd to zero and find the critical kp which you approximate to when the system oscillates.
Then you set kp, ki and kd as fractions of the oscillation period and then tune if it doesn't suit.
Logged
 

SMF spam blocked by CleanTalk