Prev: synchronize two three phase voltage supplies
Next: TOMLAB optimization
From: prerna on 30 Nov 2009 02:44
hi ,
i have a large no of digital data (3000 pts) in which there may be some spikes (not a true spike but may be called as a spike which is not defined by the relation d/dt = 0)
i want to know the algorithms for these type of spike detection
Any help would be greatly appreciated
From: Sergio on 30 Nov 2009 02:58
"prerna " <rahul8284(a)gmail.com> wrote in message <hevt47$ck$1(a)fred.mathworks.com>...
> hi ,
> i have a large no of digital data (3000 pts) in which there may be some spikes (not a true spike but may be called as a spike which is not defined by the relation d/dt = 0)
> i want to know the algorithms for these type of spike detection
> Any help would be greatly appreciated

hello,

did you try the built-in "findpeaks" routine?
From: us on 30 Nov 2009 03:08
"Sergio " <sketter2(a)zipmail.com.br> wrote in message <hevtub$je7$1(a)fred.mathworks.com>...
> "prerna " <rahul8284(a)gmail.com> wrote in message <hevt47$ck$1(a)fred.mathworks.com>...
> > hi ,
> > i have a large no of digital data (3000 pts) in which there may be some spikes (not a true spike but may be called as a spike which is not defined by the relation d/dt = 0)
> > i want to know the algorithms for these type of spike detection
> > Any help would be greatly appreciated
>
> hello,
>
> did you try the built-in "findpeaks" routine?

note to the OP:
- this requires the signal proc tbx

us
From: TideMan on 30 Nov 2009 04:39
On Nov 30, 8:44 pm, "prerna " <rahul8...(a)gmail.com> wrote:
> hi ,
> i have a large no of digital data (3000 pts) in which there may be some spikes (not a true spike but may be called as a spike which is not defined by the relation d/dt = 0)
> i want to know the algorithms for these type of spike detection
> Any help would be greatly appreciated

Try searching on despiking in Google.
You may even be lucky enough to find a paper by Goring & Nikora that
describes a number of methods.
From: Dave Robinson on 30 Nov 2009 05:00
TideMan <mulgor(a)gmail.com> wrote in message <e48f6013-1b03-47c7-bb7f-512fbd8b4625(a)h14g2000pri.googlegroups.com>...
> On Nov 30, 8:44?pm, "prerna " <rahul8...(a)gmail.com> wrote:
> > hi ,
> > i have a large no of digital data (3000 pts) in which there may be some spikes (not a true spike but may be called as a spike which is not defined by the relation d/dt = 0)
> > i want to know the algorithms for these type of spike detection
> > Any help would be greatly appreciated
>
> Try searching on despiking in Google.
> You may even be lucky enough to find a paper by Goring & Nikora that
> describes a number of methods.

What I am not sure of - are you trying to detect the spikes, or trying to remove them?

The conventional way of removing the spike is to use the median filter

help medfilt

if you have the signal processing toolbox.

If on the other hand you want the spikes, then what might work quite effectively is to subtract the median filtered waveform from the original waveform, which will generate just the spikes that the median filter has removed.

Hope that helps


Dave Robinson
 |  Next  |  Last
Pages: 1 2
Prev: synchronize two three phase voltage supplies
Next: TOMLAB optimization