It is a very simple and straight forward question.
I realy need your kind response on this...
for quadratic programming problem MATLAB has the function quadprog..i.e.,
x = quadprog(H,f,A,b)
here i can see H,f,A,b has to be known..i have gone through many literatuers but didnt find any straight forward way to find these four values form a state space system.
for example,
i have
xdot=Ax+Bu
y=Cx
and for QP i have
J=0.5xtranspose H x +ftranspose x
A x
Hi Sharmin,
H is typically a diagonal matrix that specifies how much you want to "punish" large states or state deviations from a reference in your objective function. f can be either a vector or a matrix, acting the same way like H, but in a linear and not in a quadratic fashion. A and b define inequality constraints on your states, where A is a matrix and b is a vector.
So short said: H and f must be defined by you, whereas A and b can be defined by physical limitations or desired limitations to states, for example.
I hope that was of help, good luck :-)
Dear Sharmin,
Be careful, quadprog is just a procedure to solve a STATIC problem of quadratic optimization, i.e., to minimize the quadratic functional $J=\frac{1}{2}x^{T}Hx+f^{T}x$ over the feasible set described by a set of linear inequality constraints $A\geq b$ (last inequality is understood in a componentwise or pointwise sense). Generally it has nothing common with the DYNAMICAL model in the form of state equations you have invoked. Do you wish to solve an LQ-problem for your state equations? For that you have different Matlab tools like care.m or gcare.m procedures.
Regards,
Piotr Grabowski
Dear Sharmin,
for dynamic optimization, meaning that your constraints include a dynamical system, you will have to use other tools, Piotr is right there. From my own experience, I can recommend Acado toolkit for Matlab. It is free and provides some examples, which are easy to understand, such that you can begin to implement your problem almost right away.
Good luck, Christoph
http://sourceforge.net/projects/acado/
dear Christoph,
i really appreciate your help
i have now downloaded the ACADO and will go through the tutorials..hope that it will work for my research...
one more question to u plz,if i use this tool, do i need to use MPC toolbox of MATLAB again?
sorry for frequent questions.
any help will be highly appreciated.
Hi Sharmin,
ACADO for Matlab uses its own algorithms and classes and thus, to my knowledge, does not rely on the MPC toolbox or any other toolbox related with optimization and optimal control.
Best regards
- Badges
- Science topic
- Similar topics
- Political Science
- International History and Politics
- Globalization