Hello,
I have a question. I want to develope a model Modelica using Dymola for MPC. Simulation of the MPC requires the plant simulation and the computation of the optimal solution. This is possible with JModelica? (I means it is capable to re-solve and re-implement the optimization problem).
Thanks
Yes, this is possible with JModelica.org. Please have a look at the example https://trac.jmodelica.org/browser/trunk/Python/src/jmodelica/examples/cstr_mpc.py, where you will see how the MPC controller is executed jointly with an integration algorithm to simulate execution of an MPC controller.
Best
/Johan
Hey,
Firstly, I have some questions concerning CSTR_Opt_MPC:
Maybe I don’t understand very well, in optimal control you say that the control inputs should be marked as free (optimization variables). But in CSTR_Opt_MPC I don’t see that. It’s not necessary?
Another question is why you consider for the cost
Real cost(start=0,fixed=true,initialGuess=500)
start and initialGuess?
It’s not sufficient if you have fixed=true, to consider only start=0? Or it’s just written without importance?
Thanks!
Hey, Firstly, I have some questions concerning CSTR_Opt_MPC: Maybe I don’t understand very well, in optimal control you say that the control inputs should be marked as free (optimization variables). But in CSTR_Opt_MPC I don’t see that. It’s not necessary?
The free keyword is not checked for inputs, since if there is no equation for an input it is by definition free in the optimization. For parameters on the other hand, the free attribute must be set to true in order for the parameter to be free in the optimization.
Another question is why you consider for the cost Real cost(start=0,fixed=true,initialGuess=500) start and initialGuess? It’s not sufficient if you have fixed=true, to consider only start=0? Or it’s just written without importance?
The start attribute is stated for clarity, even though setting it to zero have no effect.
Best
/Johan