Non stationary signal Processing

I will like to know how I can get the equivalent magnitude of a non
stationary signal in the time from a power spectral density plot of the
signal. 

On Apr 17, 12:28&#4294967295;pm, "ayodei" <olufem...@yahoo.com> wrote:
> I will like to know how I can get the equivalent magnitude of a non
> stationary signal in the time from a power spectral density plot of the
> signal.

So would I, let me know when you figure it out ;)

On Apr 18, 4:28&#4294967295;am, "ayodei" <olufem...@yahoo.com> wrote:
> I will like to know how I can get the equivalent magnitude of a non
> stationary signal in the time from a power spectral density plot of the
> signal.

You cannot. It's not a unique transformation back to the time-domain
and the randomness comes from an ensemble which is infinite. You can
get the autocorrelation but that's all.


Hardy

>On Apr 18, 4:28=A0am, "ayodei" <olufem...@yahoo.com> wrote:
>> I will like to know how I can get the equivalent magnitude of a non
>> stationary signal in the time from a power spectral density plot of
the
>> signal.
>
>You cannot. It's not a unique transformation back to the time-domain
>and the randomness comes from an ensemble which is infinite. You can
>get the autocorrelation but that's all.
>
>
>Hardy
>
Hardy,
Do you suggestions on how I can obtain the frequencies and corresponding
magnitudes of a non stationary signal? 

On Sat, 18 Apr 2009 01:16:53 -0500, ayodei wrote:

>>On Apr 18, 4:28=A0am, "ayodei" <olufem...@yahoo.com> wrote:
>>> I will like to know how I can get the equivalent magnitude of a non
>>> stationary signal in the time from a power spectral density plot of
> the
>>> signal.
>>
>>You cannot. It's not a unique transformation back to the time-domain and
>>the randomness comes from an ensemble which is infinite. You can get the
>>autocorrelation but that's all.
>>
>>
>>Hardy
>>
> Hardy,
> Do you suggestions on how I can obtain the frequencies and corresponding
> magnitudes of a non stationary signal?

I think the point that Hardy is trying to make is that when the signal is 
non-stationary the concepts of "frequency" and "power" (or "magnitude") 
lose a lot of their meaning, and therefore their utility as tools for 
understanding a signal's behavior as it wends it's way through any given 
system.

You _may_ be able to assign some very sensible meanings to the terms for 
the purposes of the particular problem that you are working on, but there 
is no universally sensible set of definitions for this sort of thing when 
the signal is non-stationary.

HTH.

-- 
http://www.wescottdesign.com

Please can anyone help suggest how I can analyse a non stationary signal to
get its frequency and corresponding amplitude.
>On Sat, 18 Apr 2009 01:16:53 -0500, ayodei wrote:
>
>>>On Apr 18, 4:28=A0am, "ayodei" <olufem...@yahoo.com> wrote:
>>>> I will like to know how I can get the equivalent magnitude of a non
>>>> stationary signal in the time from a power spectral density plot of
>> the
>>>> signal.
>>>
>>>You cannot. It's not a unique transformation back to the time-domain
and
>>>the randomness comes from an ensemble which is infinite. You can get
the
>>>autocorrelation but that's all.
>>>
>>>
>>>Hardy
>>>
>> Hardy,
>> Do you suggestions on how I can obtain the frequencies and
corresponding
>> magnitudes of a non stationary signal?
>
>I think the point that Hardy is trying to make is that when the signal is

>non-stationary the concepts of "frequency" and "power" (or "magnitude") 
>lose a lot of their meaning, and therefore their utility as tools for 
>understanding a signal's behavior as it wends it's way through any given

>system.
>
>You _may_ be able to assign some very sensible meanings to the terms for

>the purposes of the particular problem that you are working on, but there

>is no universally sensible set of definitions for this sort of thing when

>the signal is non-stationary.
>
>HTH.
>
>-- 
>http://www.wescottdesign.com
>

On Apr 18, 5:54&#4294967295;am, "ayodei" <olufem...@yahoo.com> wrote:
> Please can anyone help suggest how I can analyse a non stationary signal to
> get its frequency and corresponding amplitude.
>
> >On Sat, 18 Apr 2009 01:16:53 -0500, ayodei wrote:
>
> >>>On Apr 18, 4:28=A0am, "ayodei" <olufem...@yahoo.com> wrote:
> >>>> I will like to know how I can get the equivalent magnitude of a non
> >>>> stationary signal in the time from a power spectral density plot of
> >> the
> >>>> signal.
>
> >>>You cannot. It's not a unique transformation back to the time-domain
> and
> >>>the randomness comes from an ensemble which is infinite. You can get
> the
> >>>autocorrelation but that's all.
>
> >>>Hardy
>
> >> Hardy,
> >> Do you suggestions on how I can obtain the frequencies and
> corresponding
> >> magnitudes of a non stationary signal?
>
> >I think the point that Hardy is trying to make is that when the signal is
> >non-stationary the concepts of "frequency" and "power" (or "magnitude")
> >lose a lot of their meaning, and therefore their utility as tools for
> >understanding a signal's behavior as it wends it's way through any given
> >system.
>
> >You _may_ be able to assign some very sensible meanings to the terms for
> >the purposes of the particular problem that you are working on, but there
> >is no universally sensible set of definitions for this sort of thing when
> >the signal is non-stationary.
>
> >HTH.
>
> >--
> >http://www.wescottdesign.com

http://en.wikipedia.org/wiki/Spectrogram

On Apr 17, 11:28&#4294967295;am, "ayodei" <olufem...@yahoo.com> wrote:
> I will like to know how I can get the equivalent magnitude of a non
> stationary signal in the time from a power spectral density plot of the
> signal.

The Hilbert Huang Transform

(ha! I will stop discussing this now)

-Martin

On Sat, 18 Apr 2009 04:54:38 -0500, ayodei wrote:

> Please can anyone help suggest how I can analyse a non stationary signal
> to get its frequency and corresponding amplitude.

To rephrase: "Please could someone answer my question with the answer I 
want to hear?"

For that, you need to go into management.  Hire someone, tell them what 
you want to hear, tell them you'll fire them if they don't tell you what 
you want to hear, then sit back and be satisfied.

-- 
http://www.wescottdesign.com

On Sat, 18 Apr 2009 04:54:38 -0500, ayodei wrote:

> Please can anyone help suggest how I can analyse a non stationary signal
> to get its frequency and corresponding amplitude.

You're getting thoughtful answers to an ill-posed question.  Because the 
question is ill-posed and can't be answered truthfully and meaningfully 
in the form you pose it, the answers you're getting are details as to why 
the question is ill posed.

You are responding to this by asking the same damn question over and over 
again.  How in the world do you think that's going to help you?  If there 
were a good answer to your question as posed, someone would have answered 
it so far.

Instead of just whining louder, try some of the common strategies for 
correcting the problem:  say "I see that my question can't be answered, 
what question should I ask?".  Or "I think this should be answerable (and 
detail why), or "why isn't my question answerable".

Or tell us what you're trying to _do_ with the answer to the question, 
and see if someone has some positive suggestions. 

Just don't ask the same damn question a forth time.  Twice is too much, 
three times in getting irritating.

-- 
http://www.wescottdesign.com