Find Peaks tool

[jcamilleri]

Hi all,

I am new to the forum, so my apoligies if this problem has already been dicussed. I am trying to find the first 100 highest peaks from a pressure time trace which is basically a series of impact events (see Figure) using the Find Peak tool. My problem is that the algorithim is recognizing 'secondary' peaks from the same imapct event. I was thinking of first filtering the signal to remove the high frequency noise and 'secondary' peaks and then use the find peaks tool, but the magnitude of the peaks is important. So I would need to go back to the original time trace. Any ideas? 

Thank you.

Josef

'

[AndreasH_Ha]

Hello,

I looked at the picture and I am not sure which peaks are the primary ones and which are the secondary ones. Do the primary ones have a bigger amplitude ? If so, you can configure "Find Peaks" to find those first. Could you somehow mark examples for primary and secondary events in the picture you attached ?

[jcamilleri]

Hi Andreas,

Thank you for your reply. Please find attached a picture of a single impact with the peaks found by the Find Peaks tool marked. What I am calling the primary peaks are the initial highest peaks. I you need more info please let me know. 

Regards,

Josef

[AndreasH_Ha]

OK, from your pictures I would guess that you could simply search for a larger number of peaks, and then loop through those results and just keep the ones which have an amplitude larger then X. Would that work ? And if so, do you know how to do this ?

Andreas

[jcamilleri]

Hi Andreas,

Isn't that the Calculate Peaks tool? i.e., x being the lower limit value? I'm not sure how that would work but remember that the time trace is made of of a large number of single impact events like the one showed in the most recent picture, see the attached picture. So lets say the peak algorithim has found peaks A, B, C and D. When you start lopping through these results, it would discard peaks A and B and keeps peak D because it has a higher amplitude however peak D is a secondary peak occuring in impact 3. Am I right?

I was thinking more along the lines of filtering and maybe smoothing the curves, find the peaks (more importantly the time when these highest peaks occur) and then extract the peaks from the original signal? And, I may need some help with the scripting if that is possible. 

Thank you.

Josef

[AndreasH_Ha]

You are absolutely right. Now that you describe the event and the peaks I can see that what I suggested would not work. Its not just the peak. Its the peak follewing and area where the slope of the single remains more or less flat over a certain period of time. Not sure whether this is possible, but if you would share the data. I would be willing to try out a few things.

[AndreasH_Ha · Accepted solution]

Hello,

I have attached a script and a VIEW layout. The script loads the data from the folder the script is in and loads the layout from the same folder. You probably need to adjust teh filename to the name of your specific data file. The script uses some of the extensions new in DIAdem 2015, so please make sure you have this release available.

Let me know if this comes close to what you are looking for.

In VIEW you will see the area's detected by the script and the peak found in each area labled with a little symbol. I added some comments to teh script to explain what it does. Let me know if you have questions.

Andreas