SignalAnalysis

Use the component to estimate bicoherence and bispectrum from the Input signal. You can estimate bicoherence for all the frequency pairs, or just the selected ones. Bicoherence will be one for two frequency pairs, which are related in phase (like harmonics). It will be zero for frequency pairs unrelated in phase. The formula used to calculate the bicoherence:

        |E( X[f1] * X[f2] * conj(X[f1+f2]) )|
b =  -------------------------------------------
      (E(|X[f1]|) * E(|X[f2]|) * E(|X[f1+f2]|))

X.. complex frequency spectrum.
f1.. frequency index, 0 < f1 < (n-1),
f2.. frequency index, 0 < f2 < (n/2-1),
n.. length of the frequency spectrum
E.. averaging function (expected value)

Bispectrum is defined as:

B = X[f1] * X[f2] * conj(X[f1+f2])

The component features two modes of operation: Recursive and non-recursive.

• In Recursive mode (Recursive property = True),

all the spectral parameters are update for every call to the Update method. This mode can be used to run bi spectral analysis on-line in real time, if you enable infinite exponential averaging.

• In Non-recursive mode (Recursive property = false), only averaging

is performed when calls are made to the Update method. Once the averaging has finished, BiAnalyzer.Update method must be called to calculate the actual bispectra and a call to UpdateSpectrum will place the correct line from bispectrum in the Amplt/Phase property. This approach is more CPU efficient then the first. 

Because bispectral analysis is based on averaging, you can not change the parameters of the frequency spectrum of already averaged data.

Classes