Is there a LabView & Signal processing expert out there that can help with power spectrum VIs and MPF?

[gaiter]

I am trying this again.  I asked this question and may not have been clear. 

I am trying to determine the Mean Power Frequency (MPF) of an EMG signal.  I have a 10 second signal recored @1500Hz.  Using the attached VI, I calculate the MPF using a weighted average and the result is ~88 Hz is what I would expect.  But when I try to take a section of that signal (a runner's stride cycle), I am having trouble.  What I am trying to find is someone experienced in LabView AND signal processing to help me out.  I would expect a similar answer as 88 Hz (+/- 20Hz).  The 10 second signal includes about 14 stride cycles of a quadricep.

There are a number of different power spectrum VIs and I don't even know for sure if I am selecting the right one.  I have read as much LabView info as I can and still have no way of knowing if I am correct.

This is my first time trying to attached a program to a message and I selected "Save with Options" and selected "development distribution"  Hopefully this was correct.

Thanks for time.

[CKap]

Hi !!!

The problem, at least what I think of it, has nothing to do with spectral analysis. It lies in the manner you are calculating your "Mean Power Frequency".

First of all I must admit that I haven't heard of this term before.

For your MFP you are performing the following calculation

Now, if you look at my attached vi called "MPF_basic.vi" and going by your assumption, you are expecting the two outputs to be the same. But they won't be as the VI illustrates. This is mainly because of the fact that the size of the arrays for the first part of the code is 1000 and the size of the array for the second part is 100 and this affects the denominator and the numerator significantly.

Now in your original VI, the length of array for FFT power Spectral Density is 7500 and for the second one its 465 and because of this you are getting huge difference in your MPF calculations.

I would suggest using some other technique to calculate the mean frequency.

Regards,

Chetan K

Application Engineer

National Instruments

Message Edited by CKap on 11-21-2006 06:51 PM