Why have I this Result to my model with PMSM FOC with to a load in ramp ?

11 visualizzazioni (ultimi 30 giorni)
Hi,
I try to adapt the PMSM Field-Oriented Control to my application. I modelize the PMSM motor corresponding to my application and I use the block of PMSM FOC that we can have into the library (parameters adjust to my needs)
  • Simscape / Electrical / Control / PMSM Control
First question, I'm not sure but if I just want to make a current control to the motor, so I choose the "Torque" control and logically the Velocity controller is removed,that means the out loop control is opened?
Second question, is the PMSM FOC example adapted to a dynamic behavior?
I want to test the model (motor+PMSM controller) by connecting the output motor to a Torque loaded , a ramp (0 to 2Nm during 50 second) , not a constant. and I have this result (see the curves below, Torque motor and Speed motor).
I note that sometimes the speed motor is disconituous, suddenly stay constant a few moment and decrease or ring sometimes.
Someone can help me to find where I can make mistake, that can explain the result? any idea...Thanks you.

Risposte (2)

Joel Van Sickel
Joel Van Sickel il 10 Gen 2023
To your first question, making it torque control does remove one control loop. so that you are closer to current loop control, but you still have the torque control loop determining the current setpoints before the current controller, so you would have to manually remove the torque control loop as well. This can be done by right clicking the block and disabling the link to the library block under the library menu option. Then you can customize the algorithm as much as you want.
I did a search on your title "PMSM Field-Oriented Control" and was not sure exactly which example you were referring to. It looks like the velocity loop is stable as you are increasing torque. The noise you see in the response could be due to a few factors. The most likely is either the switching frequency is a bit large or the simulation parameters aren't dialled in for a really clean simulation. I wouldn't guess it is a problem with the actual controller. None of those changes in speed are actually discontinuos so while the results might not be a clean as is possible in simulation, they aren't necessarily problematic enough to need fixing. If you share the model you used to generate that plot I can take a deeper look.
  2 Commenti
Andriantsilavo Ramaroson
Andriantsilavo Ramaroson il 11 Gen 2023
Thank you to reply,
The example is this one:
I try to change my switching frequency but I have the same problem. (I drecrease this one, but I want to keep my application value, if possible)
Do you think if the PI controller parameters are not correctly adjusted, it can result this kind of response?
Another question I have, the appropriated solver is the "discontinious" with a fixed-step? and the Fundamental Sample time and controller sample time can be the same value or the controller sample time should be slower?
Must I send my model into this message or should I send it in private? (I am new in this forum so I don't really know how it works)
Joel Van Sickel
Joel Van Sickel il 12 Gen 2023
In general, it is best to attatch the model to a message in this forum, that way anyone looking at your question can look at the model. The PI values are likely fine, because when a PI controller isn't working, it's usually more obvious than the small jitter that you are seeing. I won't be able to provide more information until you share you model on this site.
For simulatin performance, ideall you should start with a variable step solver. Ode23t if you are using Simscape Electrical (Simscape core or SC) blue components, or odet23tb if you are using Specialized Technology (ST) components (the black ones). For the simscape componets you can also use DAESSC as the solver.
If you need a fixed step solver, you need to set it to at least 50 times smaller than your switching frequency, but more realistically 100 or 200 time smaller. This will effect your PWM resolution, so as Sabin mentioned below, you might be able to clean things up just by making your time step 2 to 10 times smaller than it currently is.

Accedi per commentare.


Sabin
Sabin il 10 Gen 2023
Hi,
You are correct with the assumption that when selecting ‘Torque Control’ the outer loop is open loop. We don’t measure torque and it is unusual to measure torque in real applications.
To answer the second question, we allow dynamic behaviour, an example using a ramp of torque request is the IPMSM Torque-Based Load Control example.
As the speed is quite high, the speed spikes might be related to sample time for the PWM generator block. Might be worth trying to use an averaged converter to speed up the simulation. Alternatively, if switching devices are relevant for your project, you can decrease the PWM sample time.

Prodotti


Release

R2022a

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by