I have a discrete model of a DC motor and I want to design a PID controller for it using ZN method
How can I get the PID parameters (kp, ki, kd) using ZN method?
The code is in the attached file
The method is simple, see the attachment. Even it is in Czech, the math ang algorithms are universal :)
Dear Libor
First of all, thanks for your response
However, the attached file is written with a Turkish language as far as I know
I cannot understand a single word
Secondly, The main problem in my model is that its response is fast, I cannot get the Td, which is the time from the tangent intersection to the starting.
In Czech, not Turkish :)
Sorry, I have too busy to translate the pages.
Try to use Z-N from the limit cycle test instead of the transient response, and use the corresponding tables.
I think you can get the ultimate frequency from root locus and use the Ziegler- Nichols
table and get the parameters. But ziegler- nichols controlled system gives very quick response.
I suppose the DC motor is also very quick system. May be you can design a PI controller rather than PID or can change the controller type .
Hope this helps.
Dear Hesham,
You can download the ZN method in:https://www.researchgate.net/profile/Roger_Moliner/contributions.
Furthermore your system is discrete, the ZN method is designed to continuous systems. But you can do the following can be used ZN with a continuous version of your system and then discretize the controller designed using Z transform
Dear Hesham,
With ZN method your final system will have overshoot and oscillations, which can lead to instability and cause the system to be out of control.
Maybe you should consider a different tuning method, like Internal Model Control.
https://controls.engin.umich.edu/wiki/index.php/PIDTuningClassical
Dear Hesham, your system is of second order and, therefore, the Nyquist plot does not have a point of intersection with the real axis (except the origin and zero frequency). As a result, the closed-loop ZN test would provide infinite ultimate frequency and zero ultimate amplitude. The open-loop test (step test-based tuning) would not be very helpful either because the test would not provide a response similar to the "S-curve" described in the original Ziegler and Nichols's paper. There are other methods of tuning that would be more suitable.
Actually, the ZN tuning requires assumptions and limitation to the model. Recently, I have started to use PID Tuner in MATLAB Simulnk, and it was easy to tune PID gains. Refer to the following link.
http://kr.mathworks.com/help/slcontrol/gs/automated-tuning-of-simulink-pid-controller-block.html?refresh=true
- Badges
- Science topic