System Identification Toolbox lets you estimate models from time and frequency demoing data. You can start by opening System Identification Tool and following the workflow shown by the arrows. Start this importing the data. In this case, we'll import two data sets, data t time domain data set and data f frequency domain data set.

Select the data set to work with and check the contents of this data set. If you need to, you can preprocess the data. For example, you can remove the means. You can always change the data set you work with. And if needed, you can further preprocess it. For example, by filtering the data.

Let's use the detrend and filter data set for estimating the models. And let's use a frequency domain data set, data f, for validating the results. Next, we can estimate the models. Here, we can choose from many different model structures. For example, we can estimate the transfer function.

We can estimate the state space model, the process model, and the polynomial model. We can choose different model types: ARX, ARMAX, output error, and boxed rankings. Once you have estimated several different models, you can compare them against a validation data set to see which model provides best fit.

You can also look at model residuals. You can check the time response and frequency response of your models. Look at zero pole map. You can also take the models of interest and analyze them in the LTI viewer, and you can also export the models of interest to MATLAB^® workspace for further analysis and control design. This concludes the demo.