Programatically detecting a wave from a noisy Wave.

[ank]

Dear folks,

I have a wave which i have acquired and need to figure out by means of a code what kind of wave is it. Is it a sine wave, Square Wave, Triangular wave... I did make a small program which takes the fft of the acquired wave and deconvolutes it against the fft of a standerd wave. This works fine with a pure wave and is very predictable. I then tried using the same with a realworld signal, but the noise in the wave upsetted all the value ranges. Can someone sugggest a better meathod??

Ankur

[antea06]

Hi,

you can find some suggestion in "lvmeas.pdf" in the labview manual directory: anyway you should filter the signal before doing the spectral analysis on it.

regards

[johnsold]

Ankur,

If you have only a small number of standard waveforms and they all have well-behaved mathematical descriptions, try curve-fitting the standard function to the (noisy) data and then extracting the parameters from the fitted curve.

Lynn

[ank]

antea06,

Tried that. the problem is that incase my acquired wave is essentially a square wave, the edge is distorted on filltering  and the fft of the data after deconvolution is then very close to a sine wave. So i tried the commmunities. I tried digitizing the wave and removing the lsb values of the wave. It resulted in the sine and triangular wave having very simmilar charecteristics. Thaught i could use a lock in amplifier to extract the vave, but all it gives us is a single point value in intensity of the reference wave.

Thanking You

Ankur Kumar

[antea06]

Have you already tried to averaging the signal on multiple acquisitions ?

[antea06]

I mean, in your vi (subvi Power Spectrum) what happens if you enable the averaging?