ASTDan Posted August 27, 2013 Report Share Posted August 27, 2013 Hello, I am trying to develop an algorithm that counts the amplitude of AC "events" (see attached screen shot) In the graph there are 3 events. Does anybody have any idea on how to detect this? Thanks Dan Quote Link to comment
crossrulz Posted August 27, 2013 Report Share Posted August 27, 2013 cross posted on the dark side: http://forums.ni.com/t5/LabVIEW/Algorithim-on-AC-waveform/m-p/2537348#M769268 Quote Link to comment
Jordan Kuehn Posted August 27, 2013 Report Share Posted August 27, 2013 Here is one simple implementation using a Hilbert Transform and taking a derivative of the envelope. Here's some reading that covers way more about this transform than matters for this discussion, but has a nice picture at the bottom of page 8. Something similar could be done with the ptbypt RMS suggestion you have on your dark side thread.. Take what you want from it. 1 Quote Link to comment
ShaunR Posted August 28, 2013 Report Share Posted August 28, 2013 Use the AC & DC Estimator VI to measure Vrms and just note when Vrms changes. Quote Link to comment
Jordan Kuehn Posted August 28, 2013 Report Share Posted August 28, 2013 Use the AC & DC Estimator VI to measure Vrms and just note when Vrms changes. If you went this route you could replace the hilbert function with this implementation in what I sent you and still capture the transitions in a similar fashion. Lots of options. Quote Link to comment
GregSands Posted August 28, 2013 Report Share Posted August 28, 2013 This Hilbert code doesn't cope well with changing the amplitudes (try 3, 7, 10), nor with a different first and last amplitude. Quote Link to comment
Jordan Kuehn Posted August 28, 2013 Report Share Posted August 28, 2013 Try adjusting the thresholds on the peak detection.. I debated including a graph of the derivative or not. It will behave differently than the RMS, and I'm not even suggesting it is the best option, but it can be useful. Quote Link to comment
ShaunR Posted August 28, 2013 Report Share Posted August 28, 2013 If you went this route you could replace the hilbert function with this implementation in what I sent you and still capture the transitions in a similar fashion. Lots of options. I wouldn't. I'd just take rate of change (dy/dx) of the Vrms and if it went outside a value, count the event (assuming it's a bit noisy and not zero when unchanging). Quote Link to comment
Jordan Kuehn Posted August 28, 2013 Report Share Posted August 28, 2013 That's essentially the same thing? Quote Link to comment
ASTDan Posted August 28, 2013 Author Report Share Posted August 28, 2013 I tried using a high pass filter and achieved some interesting results. I will be using real data soon, I will let everyone know the results I get. Thanks Quote Link to comment
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.