Prev: Distortion estimates & dynamical systems
Next: At the extreme of gravity there is a limit
From: Hubert on 16 May 2010 14:20
Hi,

Can you suggest any robust cycle phase extraction algorithm? I have a
signal, which contains sinus signal (assume, we know the period of
sinus). How to determine its phase? The problems are that the signal may
contain strong trends as well as gaps. The former makes it impossible to
do simple correlation with sinus signals with different phases, the
former makes it impossible (or hard at least) to do correlation of
"signal minus average over X samples" with sinus, because there are
points, where it's impossible to calculate the average (previous samples
are unknown). I'm starting to think that the second method might
actually work (the error produced by the gaps would be acceptable), but
maybe there's some other way.

thanks,
Hubert
From: Han de Bruijn on 17 May 2010 03:31
On May 16, 8:20 pm, Hubert <hub...(a)tes.pl> wrote:
> Hi,
>
> Can you suggest any robust cycle phase extraction algorithm? I have a
> signal, which contains sinus signal (assume, we know the period of
> sinus). How to determine its phase? The problems are that the signal may
> contain strong trends as well as gaps. The former makes it impossible to
> do simple correlation with sinus signals with different phases, the
> former makes it impossible (or hard at least) to do correlation of
> "signal minus average over X samples" with sinus, because there are
> points, where it's impossible to calculate the average (previous samples
> are unknown). I'm starting to think that the second method might
> actually work (the error produced by the gaps would be acceptable), but
> maybe there's some other way.
>
> thanks,
> Hubert

Isn't this just common "Fourier Analysis"? In that case, let Google be
your friend. ================

Han de Bruijn
From: Hubert on 17 May 2010 06:34
W dniu 2010-05-17 09:31, Han de Bruijn pisze:
> On May 16, 8:20 pm, Hubert<hub...(a)tes.pl> wrote:
>> Hi,
>>
>> Can you suggest any robust cycle phase extraction algorithm? I have a
>> signal, which contains sinus signal (assume, we know the period of
>> sinus). How to determine its phase? The problems are that the signal may
>> contain strong trends as well as gaps. The former makes it impossible to
>> do simple correlation with sinus signals with different phases, the
>> former makes it impossible (or hard at least) to do correlation of
>> "signal minus average over X samples" with sinus, because there are
>> points, where it's impossible to calculate the average (previous samples
>> are unknown). I'm starting to think that the second method might
>> actually work (the error produced by the gaps would be acceptable), but
>> maybe there's some other way.
>>
>> thanks,
>> Hubert
>
> Isn't this just common "Fourier Analysis"? In that case, let Google be
> your friend. ================

Not exactly. Fourier will only show harmonic frequencies depending on
the number of the samples we'll use for calculation. Anyway, i tried
calclulating covariance of differentiated signal and sinusoid of
different phases and it seems to be working fine.

Thanks anyway.
 | 
Pages: 1
Prev: Distortion estimates & dynamical systems
Next: At the extreme of gravity there is a limit