I identified the parameters of a SISO linear parameter varying simulink model by NLSE tool-box of MATLAB R2013a. One of the model parameter is a function of the previous model output, which has been represented by using a unit delay/memory block in the simulink model.
Now I'm trying to get the output from the state space model of the same system by using the model parameter identified in earlier step. I order to get output, I used 'c2d' command at Ts=2s for discretizing the continuous state space model, however, the output is not matching with the output of the simulink model. According to my best knowledge, as the model parameters are same the output should be same for the simulink and state space model but the state space model output is underdamped in comparison of the simulink model output. Then I changed Ts=1.8s (by trial & error) then it is giving quite accurate result however the sample time in both the simulation and experimental data, based on which parameter estimation has been done, were 2 sec., that's why I set Ts=2s.
My question is, why for Ts=2s the discrete state space model is not giving accurate result?
What step size you have selected in Simulink model. It mai be due to improper solver handling in simulink.
It sounds like you are doing a discrete-time simulation. Make sure you are not using a solver in Simulink and that you are using a discrete step size equal to your sampling period (set this in the config params).
Also make sure that you understand the details of the discretization of your continuous-time plant (via zero-order hold, etc). I would avoid using Matlab's ready-made functions for this and do this "by hand". You need to make sure that you understand the details and assumptions, if your matlab script is to match your Simulink script. But using both/all tools together is a very good way of making sure that everything is right.
Thank you Bharat Verma, Aparna S Murty and H L Kennedy for your responses.
- Badges
- Science topic