AC Drive China Forum

Forum » General Discussion » Flux and torque estimation for driving PMSM with DTC

Topics: **Flux and torque estimation for driving PMSM with DTC** on General Discussion
# Flux and torque estimation for driving PMSM with DTC

Can anyone help me to understand flux and torque ast,imation when using VSI FED DTC drive PMSM

#1

09-14-2013 03:27 AM

Top #2

Just Google "Flux and torque estimation for driving PMSM with DTC" for all the information you will ever need

09-14-2013 06:21 AM

Top #3

I tried it, found some papers and books, They give the same integral method, Flux = integral (V-iR)

I made simulink model acording to, the explained methods are not giving the same output with the PMSM motor in simpower module of simulink/MATLAB,

I made simulink model acording to, the explained methods are not giving the same output with the PMSM motor in simpower module of simulink/MATLAB,

09-14-2013 08:36 AM

Top #4

Torque can be expressed, in any orthogonal reference frame, for every motor, as the vector cross product of flux and stator current.

If you consider the total stator flux, this is nothing but the integral V-R*I, since the inductive and back-EMF voltage components can be all written as the time derivative of flux.

Maybe, the output of the model in Simulink is the rotor flux, which has the self-inductive flux (L*I) subtracted: flux_rotor = integral(V-R*I) - L*I ...

BTW, are you trying to implement actual DTC, which means that the switching pattern is decided just by a comparison between torque/flux reference and feedback, or are you generating inverter commands by means of a fixed-frequency PWM? In papers you can find both approaches called "DTC", even if the original one is based on comparators, which is the reason for "Direct".

If you consider the total stator flux, this is nothing but the integral V-R*I, since the inductive and back-EMF voltage components can be all written as the time derivative of flux.

Maybe, the output of the model in Simulink is the rotor flux, which has the self-inductive flux (L*I) subtracted: flux_rotor = integral(V-R*I) - L*I ...

BTW, are you trying to implement actual DTC, which means that the switching pattern is decided just by a comparison between torque/flux reference and feedback, or are you generating inverter commands by means of a fixed-frequency PWM? In papers you can find both approaches called "DTC", even if the original one is based on comparators, which is the reason for "Direct".

09-14-2013 11:15 AM

Top #5

I m trying to implement Hysteresis DTC on PMSM, I m trying to estimate flux and torque using stator referenced, not rotating (alfa,beta) reference, voltage and current values. Simulink PMSM model gives out only torque, so If my torque estimation using flux estimation is same with simulink, both estimations are correct. I think, When using ınt (V-IR) method results an offset in fluxbeta estimation occurs, so there might be a trick. I may send .mdl file prepared with Matlab R2011b

09-14-2013 01:32 PM

Top #6

You can use a filter instead of pure integration for flux estimation. The filter should be designed based on the excitation frequency. This prevents any offset in flux estimation.

09-14-2013 04:14 PM

Top #7

However, there is no reason for an offset coming out in simulation...

09-14-2013 07:04 PM

Top #8

I wouldn't use simulink PMSM model. Once you realize the differential equations yourself, you understand better what is going on. You'll have to determine if your motor is salient or non-salient one.. and other important parameters. Once you understand the Space-Vector theory you wouldn't even need the inverter (just generate the voltage vectors yourself), at least for first evaluation. however, if you do want to use the simpower blocks for the inverter, that'll be tricky (to connect the realized equations in simulink to simpower blocks).

regarding the offset - it is known in real applications that in low velocities the flux integral drifts (because not all voltage phenomena is taken under consideration when writing int(V-IR) )

regarding the offset - it is known in real applications that in low velocities the flux integral drifts (because not all voltage phenomena is taken under consideration when writing int(V-IR) )

09-14-2013 09:07 PM

Top #9

thanks to everyone,

I added the rotor flux value to the beta portion (some books call it Q) of the stationary stator flux component, now the estimated and the simulink torque output is almost similar,

I found it by trial and error, nowhere that method is mentioned

I added the rotor flux value to the beta portion (some books call it Q) of the stationary stator flux component, now the estimated and the simulink torque output is almost similar,

I found it by trial and error, nowhere that method is mentioned

05-06-2014 02:06 AM

Top #10

Can you send me your model? I am having a hard time here doing the same thing with predictive control. Thanks in advance