need guidance about how to TUNE the PI controller for exact speed tracking vector control of pmsm,i have attached the speed graph switching pulse and the model..
the problem is speed is not been tracked properly by pi controller and the D-axis goes to -300 why is that so,,what parameter should i have to check to overcome this problem
the speed controller KP=1000,Ki=5
current controller kp = 0.1, ki= 100
if in current controller i increase the kp more then 5 the speed goes to negative side and same if i increase the ki to 1000 and kp to 35 the rising time increase with more steady state error..
anyone who experience of this control technique
Hi M. Usama,
If you want helpful answers, please provide the descriptive info on the following:
1. Control block diagram,
2. State-space model of PMSM,
3. Control objectives.
the control objection for this model is field oriented control of pmsm, synchronous reference frame is used ,with svpwm switching technique we are controlling the rotor speed and position for efficient system performance
but as i change the svpwm vdc=800 and ts=0.5 so ids increase to positive and iq goes to zero or negative as mention in graph
Hi Muhammad Usama,
Since no state-space is provided, I can only make an 'educated' guess. Assuming that the state-space of the PMSM is given by:
x' = f(x) + g(x)*u,
where x are the state variables of interest, and u is the control vector.
The PMSM does not behave as desired, probably because of this part 'f(x)' is not properly compensated by your controller at a certain operating range of x.
If the exact mathematical model of PMSM is known, you can design the control laws for u, such that the closed-loop of x' = f(x) + g(x)*u is stabilized for the entire operating range of x, without 'blindly' tuning the PI controllers.
Respected sir i just used the model of pmsm that given in matlab and i am using motor sensor to find the to feedback the speed and position for current and voltage reference,i have attach the mathematical equation for the ipmsm, and i haven't design a state estimator in this model that are used in sensorless model..
Hi Muhammad Usama,
Could you point which part is x (state) and which part is u (control) in the equations that you provided?
Please transform the PMSM model into a proper state-space form.
Hi Muhammad Usama,
I don't want to burden you with the mathematical calculations. Since the dynamics of id is decoupled from iq and ω, then the control laws for ud and uq can be designed separately. Now, if such linear time-invariant state-space model is given, then the control law for uq can be designed as shown in the equations above.
With the formulas given, you can tune the gains κ1 and κ2 until the desired control objective is achieved. You may start tuning with the following initial values: κ1 = 1 and κ2 = 2. These are the fundamentals. If there is a steady-state error, then add the integral action.
respected sir Yew-Chung Chak as i attach the controller block in my previous message so can u plz guide in that regard that how can i implement these formula there!!!!!
Hi Muhammad Usama,
If you are having trouble with that, then I think it is better to use the PID Tuner assistant. It is more straightforward without the math. If you are unfamiliar with the tuning procedure, then blindly tuning control gains is like scooping the needle in the vast sea.
https://www.mathworks.com/help/slcontrol/gs/automated-tuning-of-simulink-pid-controller-block.html
Hi Muhammad Usama,
Here is a generic example of how to implement the formula to determine the control gains. In fact, the formulas are derived with the same principles of Pole Placement and Ackermann's formula.
clear all; clc
function xdot = f(x, t)
xdot = zeros(2, 1);
ref = 1;
u = 7*x(1) - 4.2*x(2) + ref/5;
xdot(1) = - 2*x(1) - 3*x(2) + u;
xdot(2) = 5*x(1) - 7*x(2);
endfunction
x0 = [0; 0];
t = linspace(0, 20, 2000);
y = lsode("f", x0, t);
plot(t, y);
legend('i_{q}', '\omega_{r}');
xlabel('Time (sec)')
- Badges
- Science topic
- Similar topics
- Geoscience
- Asia
- India