I am making a simulink model for three phase grid tied inverter. I have a few questions. Can you help me out?
-At the summing point (Id-Iq*wl*Vd) and (Iq+Id*wl*Vq) I am getting very high value that sets Vabc ref for pulse generation very high value. How to do scaling?
- How to set pi controller parameter? Is there any specific approach to find those values?
Thank you in advance
1.- The ouput of the current control loop is the voltage reference for PWM block. Therefore, it should be in volts and its amplitude should be of the same order as the grid voltage. Of course, for setting the PWM references, you should normalize the voltage with the dc-voltage.
2.- The Kp of the current controller is a function of the output filter inductance and intended bandwidht. The maximum bandwidth is a function of sampling frequency, and is usually limited to 1/10 fs. In practice, something in the range of 0.07 fs is good. An example:
Kp=L*(0.07 ws), with ws=2pi fs (e.g. fs=5kHz).
the Ki is a function of R,
Ki=R*(0.07 ws)
but this R should also include the conduction losses (not only the filter resistance, that usually is very small).
By proper tuning of PI controller, you can get the right values of Vabc.
The parameter of PI controller can be get by try error, but, you can get an educated guess of right values by Ziegler-Nichols.
1.- The ouput of the current control loop is the voltage reference for PWM block. Therefore, it should be in volts and its amplitude should be of the same order as the grid voltage. Of course, for setting the PWM references, you should normalize the voltage with the dc-voltage.
2.- The Kp of the current controller is a function of the output filter inductance and intended bandwidht. The maximum bandwidth is a function of sampling frequency, and is usually limited to 1/10 fs. In practice, something in the range of 0.07 fs is good. An example:
Kp=L*(0.07 ws), with ws=2pi fs (e.g. fs=5kHz).
the Ki is a function of R,
Ki=R*(0.07 ws)
but this R should also include the conduction losses (not only the filter resistance, that usually is very small).
I would add to Francisco awnser that adding saturation to your PI controller will help when it comes to fine tuning the PI.
The output of the PI, which is a current, should not be allowed to be bigger than 4 times the nominal current.
From my experience, when there is saturation, no matter the PI gain, the model find a reference, eventually.
- Badges
- Science topic