In my paper for design of loop filter instead of PID loop filter approach, I employed modern predictive control (MPC) in accordance to attached file. But reviewer of my paper declare “the novelty of the paper is tested by MATLAB simulator does not have enough accuracy. I believe that the novelty of the paper should be tested by a better and more accurate simulator”.
As far as I know, all the papers similar to my work used MATLAB for simulation. I appreciate it if anyone would propose a solution for this problem.
Hello!
Matlab has a huge variety of solvers for both continuous time or discrete time...you should choose the method that best suits, but usually (if your system is non stiff or marginally stable) the ones with less computational effort do well!
See: http://www.mathworks.com/help/matlab/math/choose-an-ode-solver.html
First, I suggest that you simulate your system twice with two different Simulation methods just to be sure that "the simulation isn't completely wrong".
Second, if you need more accuracy, try to
-DECREASE the step size for discrete time systems (which is probably your case if you are dealing with Model Predictive Control)
-DECREASE the tolerance for continuous time Systems
Hope it was helpful
Try simulating with Labview . It is faster, it has shorter processing times . In precision it would be good to try .
In matlab you can also cambair accuracy, with command options optimizers .
Dear Mohamad you can adjust the accuracy of your matlab simulation as you wish. You can choose the solver and whether to use variable steps (faster) or fixed steps that you can set but on the cost of increasing simulation time. use very small step size can generate very accurate results similar to the continuous ones.
- Badges
- Science topic