Thank you for this explanatory video, but the example shown is chosen very wrong. You should never tune the controller for the wheels with the wheels raised in the air. When the wheels touch the ground so the motors are under load, the system dynamics (and also the mathematical model) will be completely different. So the controller you set while the wheels are in the air will not work as you wish.
I have a question. When I try to use my Simscape model, I can't close the loop directly because it would create an algebraic loop and the simulation would have unpredictable behaviour and discontinuities. To solve this, I added an almost transparent first order transfer function, to remove the loop. However, by implementing that solution, Simulink does not recognize the Control Loop, and it can't autotune the PID controller. Any ideas?
Hi, I'm looking for a way to upload FC drone blackbox reports as input and the sfw give me a PID value for roll pitch and yaw that minimize the PID error. Could help me?
Hello, I wanted to tune a discrete PID controller using GA, how can I do that? Can anyone please help me , by sharing the code or what to write in the code (i work in the z domaine "discret ")
Matlab is just too expensive unless you're a student or work for NASA, Boeing or another huge corporation. The software looks open source at first glance but the costs to use are incredible.
It's a business and MathWorks is not Santa Claus. Google or Facebook can opensource their APIs, but they earn via advertisements and other paid services. How do you think that MathWorks is gonna earn if they give their products for free as in the case of Google and Facebook? Advertising? Where? How? Stop whining! Use the free Octave (Matlab alternative) if you don't want to pay for Matlab license and see how far you can go in being productive.
Dear MATLAB, What do you mean by "automatic tuner"?! Assigning gains of the controller is very subjective and it depends on what the user wants in terms of speed, overshoot, level of disturbance rejection, robustness, etc. Now, how does your "automatic tuner" decide about that for the user?! How does it figure out the user's personal desires for the controller?
Did you ever opened this feature? I am asking because it is possible to tune response (thus tune controller gains) to desired speed, overshoot and other, even robustness. It is really good feature to obtain gains. Actually, setting controller response is nothing special for linear system, and for some linearized nonlinear systems in operating range (if there is no large nonlinearity). Also, this feature can also solve integrator windup problem, and in my practical uses it works fairly good. Certainly better than setting it "manually" or using some methods as Zieger-Nichols. As you maybe know, there are tuning methods that guarantee aperiodic response, such as damping optimum method, so it is not a big deal. Proven in practice for more than 50-60 yrs. Finally, a good PID (in my experience I+PD form) really solves more than 90% of all control problems.
@@mkrznar Then it wouldn't be automatic! It is still YOU who is declaring your desired performance parameters not this "automatic tuner". As for the PID, I think state feedback a stronger tool as it gives you more mathematical tools to make your system optimal, robust, adaptive, etc.
There is no point to go to state feedback for simple problems such as presented herein. It is overkill. Further more, there is a vast class of problems where it is not possible to have mesurements nesecery for state feedback, and you need observers then. This is also overkill sometimes and model dependent. This toolbox is automatic, because you get good gains by clicking tune button (and stable response). If you want to refine controller you can add additional constrains that are available inside this toolbox. Do not forget that all autotuning methods depend on model/indentification accuracy, and from your point of view there is no automatic pid tuner.
@@mkrznar Thanks for the explanation. The problem I have now is I have controlled a system that is extremely nonlinear with a PD controller but I can't optimize the gains in terms of speed and energy consumption of the controller without doing trial and error manually. How do you think this can be performed on a nonlinear system?
Thanks for sharing MathWorks,I hope more new content about PID
Thank you for this explanatory video, but the example shown is chosen very wrong. You should never tune the controller for the wheels with the wheels raised in the air. When the wheels touch the ground so the motors are under load, the system dynamics (and also the mathematical model) will be completely different. So the controller you set while the wheels are in the air will not work as you wish.
For non liner system it is difficult to find transfer function. Then how to use auto tuning
I have a question. When I try to use my Simscape model, I can't close the loop directly because it would create an algebraic loop and the simulation would have unpredictable behaviour and discontinuities. To solve this, I added an almost transparent first order transfer function, to remove the loop. However, by implementing that solution, Simulink does not recognize the Control Loop, and it can't autotune the PID controller.
Any ideas?
Use a unit delay 1/z in the position feedback loop
How to select transfer function of a system?
How to tune, when it is used in inverter?
Hi, I'm looking for a way to upload FC drone blackbox reports as input and the sfw give me a PID value for roll pitch and yaw that minimize the PID error. Could help me?
Is this tuning process works for non unity feedback systems as well?
Hello, I wanted to tune a discrete PID controller using GA, how can I do that? Can anyone please help me , by sharing the code or what to write in the code (i work in the z domaine "discret ")
Thank you for sharing.
Selamün aleyküm kral sana bişey danışabilir miyim
Thank you very much
This is linear system, how to PID for non-linear system?
I've made a monte-carlo solver for such task. Took a plant out of use for a week, but worked like a charm.
why dont you show the control signal? just making an aggressive controller is simple, the whole art is making the control signal smooth
Thanks
Thanx alot engineer
Matlab is just too expensive unless you're a student or work for NASA, Boeing or another huge corporation. The software looks open source at first glance but the costs to use are incredible.
It's a business and MathWorks is not Santa Claus. Google or Facebook can opensource their APIs, but they earn via advertisements and other paid services. How do you think that MathWorks is gonna earn if they give their products for free as in the case of Google and Facebook? Advertising? Where? How? Stop whining! Use the free Octave (Matlab alternative) if you don't want to pay for Matlab license and see how far you can go in being productive.
Dear MATLAB,
What do you mean by "automatic tuner"?! Assigning gains of the controller is very subjective and it depends on what the user wants in terms of speed, overshoot, level of disturbance rejection, robustness, etc. Now, how does your "automatic tuner" decide about that for the user?! How does it figure out the user's personal desires for the controller?
Did you ever opened this feature? I am asking because it is possible to tune response (thus tune controller gains) to desired speed, overshoot and other, even robustness. It is really good feature to obtain gains. Actually, setting controller response is nothing special for linear system, and for some linearized nonlinear systems in operating range (if there is no large nonlinearity).
Also, this feature can also solve integrator windup problem, and in my practical uses it works fairly good. Certainly better than setting it "manually" or using some methods as Zieger-Nichols.
As you maybe know, there are tuning methods that guarantee aperiodic response, such as damping optimum method, so it is not a big deal. Proven in practice for more than 50-60 yrs. Finally, a good PID (in my experience I+PD form) really solves more than 90% of all control problems.
@@mkrznar Then it wouldn't be automatic! It is still YOU who is declaring your desired performance parameters not this "automatic tuner". As for the PID, I think state feedback a stronger tool as it gives you more mathematical tools to make your system optimal, robust, adaptive, etc.
There is no point to go to state feedback for simple problems such as presented herein. It is overkill. Further more, there is a vast class of problems where it is not possible to have mesurements nesecery for state feedback, and you need observers then. This is also overkill sometimes and model dependent. This toolbox is automatic, because you get good gains by clicking tune button (and stable response). If you want to refine controller you can add additional constrains that are available inside this toolbox. Do not forget that all autotuning methods depend on model/indentification accuracy, and from your point of view there is no automatic pid tuner.
@@mkrznar Thanks for the explanation. The problem I have now is I have controlled a system that is extremely nonlinear with a PD controller but I can't optimize the gains in terms of speed and energy consumption of the controller without doing trial and error manually. How do you think this can be performed on a nonlinear system?
@@SaeedAcronia when I needed an autotune PID for a nonlinear model, I just made an autotuner myself via the critera I need for my exact case.