Spectral features provide frequency-domain metrics on your data. To compute spectral features, you must already have a power spectrum or an order spectrum variable.
Spectrum — Choose from the available spectrum variables. The software brings up the plot of that variable for reference, and converts the plot from log scale to linear scale.
Peak amplitude — Generate a feature based on the amplitude of the peaks.
Peak frequency — Generate a feature based on the frequency of the peaks.
Peak value lower threshold — Constrain peak size to exclude
low-amplitude peaks. For more information, in findpeaks, see the
MinPeakHeight name-value pair argument.
Number of peaks — Number of peaks to generate features for.
The software selects N most prominent peaks in the chosen frequency band, going in the
descending amplitude order. For more information, in findpeaks,
see the NPeaks name-value pair argument.
Minimum frequency gap — Specify a minimum frequency gap. If
the gap between two peaks is less than this specification, the software ignores the
smaller peak of the pair. For more information, in findpeaks, see
the MinPeakDistance name-value pair argument.
Peak excursion tolerance — Specify the minimum prominence of
a peak. The prominence of a peak measures how much the peak stands out due to its
intrinsic height and its location relative to other peaks. For more information, in
findpeaks, see the MinPeakProminence
name-value pair argument.
For more information, see findpeaks.
Modal coefficient features extract modal parameters from frequency response information. The setting for Number of peaks determines the number of modal coefficients to return.
Natural frequency — Natural frequencies corresponding to each mode
Damping factor — Damping factor for each mode
For more information, see modalfit.
Band power represents the power of the signal in the selected frequency band. It is defined as the area under the spectrum curve within the chosen band limits.
Use the slider to set the frequency range within which the spectral features are computed. The slider settings interact with the spectrum plot by shading the active area, effectively providing a data cursor. As you move the slider, the edge of the active area in the plot moves correspondingly.
The software stores the results of the computation in new features. The new feature
names include the source signal name with the suffix spec.