I designed a robust controller based on H_infinity methods with help of matlab, I try to implemented this controller in hardware like (arduino,raspberry pi,pcduino...),is there an example?
or some thing can help to do that?
Hi Khaled,
First, discretize the controller for a given sampling time (Ts). You can then obtain the controller in state-space form (A,B,C,D matrices). Inside your timer interrupt, which gets executed every Ts time interval, you may write the difference equations:
x_new = x + (A*x + B*u)*Ts;
y = C*x_new + D*u;
x = x_new;
Here, u is the input to controller and y is the output of controller.
Hope this helps!
If you really want to implement some analog controllers, there is a relatively newer technology (FPAA), which lets you design continuous-time controllers.
Hi.
If you search for "raspberry control system" you will find at least one supplier with a free evaluation package (free editor, 2 hours run-time) for control implementations.
The package I found was very complete with both modelling and control libraries.
The most delicate phase in your project it is the implementation part using a specialized processor of your choice. For this I suggest you study the documents whose links are:
- radio-hobby.org/uploads/journal/.../cc_12_13.pdf
- simpson.net.au/david/292.pdf
- tronixstuff.com/feed/
Good Luck Khaled
Dear Khaled,
Check the REX control system which may be suitable for your application.
https://www.rexcontrols.com/rex-control-system-raspberry-pi
It provides a real-time core and an extensive library of functional blocks which may be used to design your control algorithms without a necessity of hand coding (just like in the Matlab-Simulink environment). There are development tools which you may try for free.
Various target hardware platforms are supported including Raspberry Pi. You can connect your system to the controlled plant via proper inputs/outputs in numerous ways (GPIOs, expansion boards or external devices supporting standard communication protocols).
Regards,
Martin
Hello Khaled,
In general I agree with Mayank Baranwal, however, I'm not sure about this line:
x_new = x + (A*x + B*u)*Ts;
Since {A,B,C,D} are already discrete state space matrices, it seems that the above line should be replaced by:
x_new = A*x + B*u;
Best regards,
Sergei
hinf state space matrices are
(A, B1, B2, C1, C2, D11, D12, D22, D21)
the problem in loophapping is who can I make code for the compensatore W1 W2 W3 .. ????
Dear Khaled,
If you are confused about the selection of appropriate weighting functions w1 w2,w3, type in MATLAB help "Robust control toolbox examples" and just follow the relevant example. Notice that w1,w2,w3 are used for the controller design only, they are not implemented.
Best regards,
Sergei
Hi Khaled,
I agree with Sergei and here is the documentation link for explaining the entire procedure with examples.
Compute H∞ optimal controller for LTI plant - MATLAB hinfsyn
http://www.mathworks.com/help/robust/ref/hinfsyn.html
Good Luck
- Badges
- Science topic
- Similar topics
- Engineering
- Control Theory
- Control Systems
- Control Systems Engineering