3
$\begingroup$

I am pretty new in the field of Signal Processing. I am a structural engineering graduate student at the UT at Austin. I am currently working on a project that regards acoustic emission monitoring of a steel-concrete composite bridge.

I am analyzing some Power Density Spectrums that I calculated from the waveforms collected during the experimental program. Now, I am trying to classify some of them basing on Mean Frequency and occupied bandwidth. I have two main questions :

  1. Could someone gently suggest some literature reviews where I can read more about this topic?
  2. My actual classification method involves the inverse of the occupied bandwidth, so 1/obw. Does it have any meaning in the field? I mean, is it a feature that could be find anywhere? I also need some literature regarding it.
$\endgroup$
  • 1
    $\begingroup$ The inverse of a frequency is time, and thus, you could interpret the inverse of a bandwidth as "time certainty". $\endgroup$ – Marcus Müller Nov 5 '16 at 23:59
  • $\begingroup$ Thanks for your reply. Also, do you think that applying Gabor Uncertainty Principle to have a sort of mathematical prove of the concept could be reasonable ? $\endgroup$ – Marco Manconi Nov 7 '16 at 0:41
  • $\begingroup$ Dear @Marco Manconi: Do you have any clear procedure on how the accoustic excitations were introduced?, Forced? Natural? Are some source receiver testing involved? Of course the badwidth has a clear meaning on this kind of experiments. Why are you forcing introducing a so vague concept (dtdf ratio) in here? The dfdt ratio only is applicable when the sampling time (dt) and the time length (df) are insufficient: meaning you did all the job bad... $\endgroup$ – Brethlosze Nov 12 '16 at 8:32
  • $\begingroup$ i haven't been too familiar with the term "occupied bandwidth", but i might guess that Q (for quality factor). there are well-known relationships between Q and bandwidth (either in Hz or in octaves) for 2nd-order IIR filters. that concept can be extended from filter spectrum to signal spectrum. other than that, i dunno what would be the relationship. $\endgroup$ – robert bristow-johnson Jan 6 '17 at 1:33
1
$\begingroup$

(1) Here is one interesting reference on vibration spectral analysis.

(2) Inverse of occupied bandwidth could correspond to a relaxation time of some vibration-- the time it takes the oscillations to die down to a negligible amplitude.

$\endgroup$
  • $\begingroup$ Thanks for your reply. Also, do you think that applying Gabor Uncertainty Principle to have a sort of mathematical prove of the concept could be reasonable ? $\endgroup$ – Marco Manconi Nov 7 '16 at 0:41

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.