How I can system Identification by Matlab/Simulink and what is the best reference?
You can use adaptive filters, LMS algorithm for identification of linear systems. There are already toolboxes available in MATLAB for this purpose.
http://www.mathworks.com/help/dsp/ug/lms-adaptive-filters.html
You can also use system identification toolbox in MATLAB, which allows you to fit a model to your input-output data.
http://www.mathworks.com/products/sysid/
Maybe you should look at the following books: 1. Ljung System Identification; 2. Isermann; 3. Peter Young, and so on. In some of these books the matlab identification and simulation procedures are also explained.
if your identification is in frequency domain a simple method is FFt (Fast Fourier transform ) .
you can use FFt function or block in MATLAB
this method can give the natural frequencies of signal
wavelet method is used when the gradient of a signal is very intense
in this case ,you can use wavelet toolbox in MATLAB
Try ident
System Identification Tool provides an interactive environment for
data analysis, model estimation and response visualization.
Using this GUI, a variety of models (non-parametric, and linear/nonlinear
parametric) can be estimated and their responses visualized and compared
to each other.
The GUI facilitates estimation of both continuous-time and
discrete-time models, using both time and frequency-domain data. Start by
opening the GUI and importing data sets.
The appropriate identification method depends on the identification problem (process), which you are trying to solve (identify). In the case of parametric identification, the structure (order) of the model and its parameters must be determined (estimated). It is also important if the identified model of the process will be used for the analysis purposis, prediction purposes, in the context of process control, and so on. If you only observe certain time series (generated by certain stochastic process), then you should maybe try to use the Box Jenkins family of models (for example ARMA or ARIMA). If there are also exogenous inputs involved, which influence on the observed proces, the use of ARMAX or ARIMAX models might be the right choice. Besides the identification toolbox, there are also other toolboxes possible to use, such as: NNSID toolbox, econometric toolbox, and so on.
Farzin:
I've used the system identification toolbox for identifying linear state space models using subspace methods. It is very easy to use and the models tend to be very accurate. I can help you with any specific questions on the subject.
dear,
you can use the command "ident" in the workspace of matlab, a Box will be appeared which is very easy to use.
In nonlinear systems you should use identification techniques nonparametric , which involves reviewing the data obtained from the samples, you apply a moving average filter to cancel out noise so you can use the GUI identification with ARMAX technique that can give you more than 90% of effectiveness .
The book Ljung Identification System has the steps .
- Badges
- Science topic