identifying continuous-time and discrete-time models using ident toolbox

Question

Mohammad 2017-4-18

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/335980-identifying-continuous-time-and-discrete-time-models-using-ident-toolbox

评论： Mohammad 2017-4-18

Hello,

I have some questions regarding system identification toolbox.

I am running a simulink model and collect data with "to workspace" block as my inputs and outputs to be used in ident toolbox.

1- What sample time -1 means for "to workspace" block? If I want to identify a continuous-time model in ident toolbox, should I set that sampling time to -1?

2- If I want to identify a discrete-time model can I set sampling time of those blocks to, say 0.1, and then set the sampling time field in ident toolbox on 1, and choose discrete-time option? Could you please tell me where that disturbance component (signal) is exactly located in the closed loop system?

BTW, if I choose discrete-time option, then the state-space form will be x(t+Ts)=..., if I set Ts=1 is that equal to x[k+1]?

Bests, Mohammad

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Arkadiy Turevskiy 2017-4-18

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/335980-identifying-continuous-time-and-discrete-time-models-using-ident-toolbox#answer_263501

Hi there,

Please take a look at this example , I think it should help clarify things.

Sample time of -1 means the sample time will be inherited from the upstream part of the model. For purposes of system identification data has to be sample uniformly, so instead of -1 specify a small sample time you want to sample simulation data with. Then follow the workflow described in the model for estimating either discrete or continuous model.

If you estimate discrete transfer function or state space system the equation will be something like x(k+1)= function (x(k), u(k). Here k+1 refers to the next computation after k. Whether sampling time is 1 sec, 0.1 sec or 0.01 sec, the discrete time system is always described by the equation I mentioned.

Hope this helps.

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Mohammad 2017-4-18

Hi Arkadiy,

Thanks for your answer.

I think there is still one thing unclear for me:

How can I estimate a continuous-time state-space model for my data? I am pretty sure that I have to choose "continuous-time" option in ident toolbox. But the thing is that should I first reset the sample time of "to workspace" blocks from, say 0.1 sec, to -1?

Bests

请先登录，再进行评论。