Count number of peak-to-trough excursions above a threshold

Hi,

I am hoping someone in the Forum can help me with either either cell formulas or VBA code to sort out some data peaks. Attached is a worksheet containing spectral data. The original plan was to identify when an event was ocurring by filtering out low frequency data and only observing spectral content above 14kHz. It worked in the lab however, in the real world, there was noise in the same band as the signal and at a higher level.

When I plotted the data by subtracting the event and noise spectrums from a reference spectrum (no event or noise present), I found a pattern that could identify the event compared to the noise. The event has more pronounced peak-to-trough excursions compared with the noise. Unfortunately, testing shows that while the greater peak-to-trough excursions are consistent when the event occurs, the exact spectral peaks is not.

What I was thinking might work is: at frequencies above a designated cutoff, count the number of peaks detected above a designated peak-to-trough excursion value.

In cases where the troughs on either side of the peak are not symmetrical, the peak to trough value would be calculated from the peak to the greater of the two trough values.

If anyone in the Forum can help it would be greatly appreciated.

Thanks,

Art

After putting up my post, I found the following link that might be helpful to anybody trying to work on the problem that I posted:

http://stackoverflow.com/questions/8...graph-analysis

The sumproduct formula finds the peaks, but does not qualify them as being above a threshold. Hopefully, this will be directional in solving that problem.

Thanks,

Art