How to turn a continuous time-y(t) array into Zero-Order-Hold and Tustin method discrete plot with different Ts

9 次查看（过去 30 天）

Pablo Graale 2024-4-12

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/2105891-how-to-turn-a-continuous-time-y-t-array-into-zero-order-hold-and-tustin-method-discrete-plot-with-d

评论： Pablo Graale 2024-4-12

Hi, I've obtained the continuous plant of a cone-shaped-tank on Simulink. I took the data and turned into arrays both for time and y(t). I'd like to discretize both with ZOH and Tustin.

I haven't found any method yet.

Thank you in advance,

Pablo.

4 个评论
显示 2更早的评论隐藏 2更早的评论

Manikanta Aditya 2024-4-12

在 MATLAB Online 中打开

If you have time-series data and you want to estimate a transfer function, you can use the tfest function in MATLAB’s System Identification Toolbox. Here is a simple example:

% Assuming 'time' and 'y' are your arrays
data = iddata(y,[],Ts); % Create an iddata object, Ts is your sampling time
np = 2; % Number of poles
nz = 0; % Number of zeros
sys = tfest(data,np,nz); % Estimate transfer function

Once you have the transfer function, you can then discretize it using the Zero-Order Hold (ZOH) and Tustin methods as mentioned in the previous comments.

% For Zero-Order Hold
sys_zoh = c2d(sys, Ts, 'zoh');
% For Tustin method
sys_tustin = c2d(sys, Ts, 'tustin');

Pablo Graale 2024-4-12

Mmm it doesn't allow me to use the tfest function with the data obtained with the iddata command. I've obtained something replacing the [] with the time, but I'm not certain the obtained data makes sense.

With the data from my system on Simulink I had the following

What the second plot means, is that the cone-shaped-cone takes about 25 seconds to empty and the dashed line is the linearized system I obtained with the Taylor expansion. The system looks a bit complex, I wish it was just a transfer function