Correlation Models

The Correlation Model dialog is opened by choosing the corresponding item in the pop-up menu Estimate in the System Identification app's main window.

This function estimates the impulse response of the Working Data set using high order FIR models (closely related to correlation analysis). The result can be viewed by opening the Transient resp model View.

The calculation is also obtained by pressing the hotkey c in the window.

For the calculations, the input is first prewhitened using an AR filter. The order of this filter (default 10) can chosen as an option in the dialog box.

The algorithm is described in more detail under IMPULSE in the manual.

The Time Span over which the impulse response is calculated can be entered in the corresponding edit box. Negative lags may be useful to investigate any feedback effects in the input-output data.

The default time span is based on the number entered under 'Time span' in the 'Options' pull-down menu in the 'Transient Response' Model View window.

NOTE: This function cannot be applied to time series data (no input).

Help topics.

Pre-whitening for Correlation Model

Pre-whitening for Correlation Model

The pre-whitening filter is of the form A(q)u(t)=u_F(t), i.e., the input u(t) (and output y(t)) is subjected to an FIR filter A to produce the filtered signal u_F(t). The order of the pre-whitening filter, N, is the order (number of lags) of this filter A. The filter is determined by modeling the input as an AR process of order N. The default value of N is 10.

This default value is also obtained by leaving the corresponding edit box empty.

Help topics.

(file cra.htm)