Most of dynamical systems are controllable by Model Predictive-based controllers. You need to specify what kind of approach you want to follow: Averaged state space model, hybrid model...
It seems it is possible in theory, since it has been done (at least in simulations, i dont know if a hardware+software implementation exists). You can e.g. find a comparison between fuzzy & model predictiv controlled buck converter here:
http://ieeexplore.ieee.org/xpl/articleDetails.jsp?tp=&arnumber=7041979&queryText%3DBUCK+CONVERTER
About your second question: "How to make a simulation of an MPC -buck converter?" i assume you mean to implement receiding horizon MPC?
=> you 'll need an implicit or explicit emulator model (the buck converter), an explicit controller model (often less detailed/accurate than the emulator model), a target trajectory, predictions for all disturbances and a chosen control timestep and control horizon.
Once you have these, you can use your modeling or simulation software to simulate and apply the predictive controllers output on the emulator model.
E.g. in matlab, there is the model predictive controller toolbox, and sim() function. Or you could use simulink.
In fact, Fuzzy and MPC are kindly "heavy implementable" algorithmes for such converter. I think that an optimal-tuned PID controller is enough in terms of implementation (actually, analog PI can be enough). If you have a system that includes the buck converter with any other system (battery, loads...), maybe the MPC is a possibility.
One last opinion: when you have knowledge of your model you don't need to implement a fuzzy-based strategy, because you already know the model, and you can also define the performance. Of course, you can prevent saturations with "intelligent" controllers, but this can be also achieved with any class of constrained strategies. With the MPC you need a good model knowledge.
I agree with John on this.
MPC will be typically be economically un-justified overkill for this case, potentially increasing the efficiency slightly, but at the indirect cost of a large increment in complexity and engineering time.
The opposite is also possible: a well tuned PID controller can outperform an MPC in noisy or uncertain conditions.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Control Theory
- Control Systems