Re: frequency analysis of long transient signal


<richard2008wang@xxxxxxxxx> wrote in message
news:1170085828.196590.220970@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
Hi, everyone.
I must admit my naiveness in signal processing, so really appreciate
your any possible help.

I have a long time transient signal, i.e. it takes a quite a long time
to decay to zero. My aim is to get the frequency spectrum of this
signal as accurately as possible. But now I can only obtain a limited
period of this signal from the start time. And all the material is
only this limited period of signal.

So is there any advanced technique which can help me to work out the
frequency spectrum of the whole signal to a relatively high precision
by only analyzing this short period signal?

I totally have no idea, so any help or any tip is appreciated.


No. By definition a transient is a signal that is not part of the long term
behavor(which is usually periodic or quasi-periodic).

If you sample only the first 10 seconds then you have no idea what happens
in the next 10 and the frequency spectrum could be totally different.

I think this sorta might have to do with causality too if I understand your
problem. Suppose your tansient is a sinc function. Then your frequency
spectrum is infinite and no matter how long you sample you'll never get it
all.

So what you really need to do is look at your problem more and find some
criteria that can help you reduce your problem to something thats workable.

I might not be understand what you want though. But suppose we have a a
signal we are trying to measure such as an impulse that might happen at any
time between 0 and 10 seconds. If we just look at the first 5 seconds then
there is no way we can know when the impulse occurs if it doesn't occur in
the first 5 seconds. In this case though we do know the spectrum because we
know the signal. In general though its impossible if you don't have some
idea what is going to happen.

By your logic we could just sample for 1 second and be able to know the
spectrum of all future events no matter what happens.





.

Relevant Pages

  • Re: frequency analysis of long transient signal
    ... to decay to zero. ... My aim is to get the frequency spectrum of this ... only this limited period of signal. ... signal we are trying to measure such as an impulse that might happen at any ...
    (sci.math)
  • Re: frequency analysis of long transient signal
    ... My aim is to get the frequency spectrum of this ... spectrum is infinite and no matter how long you sample you'll never get it ... signal we are trying to measure such as an impulse that might happen at any ...
    (sci.math)
  • Re: Lightning and Fourier transform of an impulse
    ... So I don't think this can be used to explain why the lightning ... enough to localize the interference in time. ... An impulse creates all frequencies, and they exist for all time. ... The frequency spectrum of ANY signal exists for all time. ...
    (comp.dsp)
  • frequency analysis of long transient signal
    ... I must admit my naiveness in signal processing, ... My aim is to get the frequency spectrum of this ... only this limited period of signal. ...
    (sci.math)
  • Re: Lightning and Fourier transform of an impulse
    ... The lightning isn't really an impulse. ... So the Prof can't use the broadband nature of the frequency spectrum ... He can say the spectrum is broadband. ...
    (comp.dsp)