LabVIEW

cancel
Showing results for 
Search instead for 
Did you mean: 

Smart peak detection

It seems that the reason why you are unable to find the peaks with the "Waveform peak detection.vi" is because there is also another signal with a lower frequency that is pushing the desired signal up and down. Instead of a low-pass, try a band pass filter. I attached an example for you to play with. Note that the band-pass gives you a singal that is better for the peak detector.

Now, be aware that I don't know much about your signal, so this may be dead wrong.

The other problem you need to figure out is how/when to change the cut-off frequencies as the acceleration change.
www.vartortech.com
Message 11 of 16
(1,206 Views)
I forgot to mention that the resulting signal is out of phase. You need to compensate for that. Also, I assumed a 1kHz sampling rate.

Mensaje editado por Enrique

www.vartortech.com
0 Kudos
Message 12 of 16
(1,205 Views)
small deviations = I'm sorry but I can't find clear words to describe what I mean. Picture will be better - violet signs are "deviations" because these peaks are on one big peak - red sign. Please tell me the right word for these regions (bunch of bumps?) I think they can be "erased" with averaging.

Picture 5 was my mystake. The peak around 130 is not interesting for me 160 as well, so I didn't ring. But this is still the type of peak I'm looking for. But as I wrote I'm interested only in the third period of the signal (this is part where the valve "landed")

In addition this is only one curve - I have hundreds of them and the algorithm must work for all of them or at least 80%

I think good way should be resampling, averiging and then use peak analyzer...
LV 2011, Win7
0 Kudos
Message 13 of 16
(1,198 Views)
 
LV 2011, Win7
0 Kudos
Message 14 of 16
(1,192 Views)
Hi David,
I had problem with property node, there is "cursor.cursor Y" but in LV7.1 there is not this property. Is it similar to Cursor.PosY? It works with it but it doesn't eliminate small bumps.
LV 2011, Win7
0 Kudos
Message 15 of 16
(1,190 Views)

my apologies ceties

I hadn't realised that that the VI would break downconverting from LV8.0 to LV7.1. Yes that the correct cursor. I have fixed the cursor issue and attached the VI again. The VI isn't perfect for your dataset. All I wanted to do was suggest that your starting point could be one of the shipped examples.

David

0 Kudos
Message 16 of 16
(1,164 Views)