1
$\begingroup$

Iam using acoustic waves to measure the thickness of material. My problem is that the acoustical lense produces static echos in the signal. I try to seperate both because there is a short interference between both.

The frequency of the material echo and the static echo has neary the same frequency. So spectral subtraction is not a valid method to sperate the signals.

I will try the following: The first measurement is without any object. So iam able to measure the static echo of the acoustic system. The second mesurement i collect the signal with the object to measure.

My idea: So i got two signals from different signal sources. Maybe i can seperate both with Independet Component ANylysis (ICA).

Do you think that this solution will work? Or is there a better method to seperate static and echo signals (with noise) out the measured data.

Thanks for your help.

$\endgroup$
  • 1
    $\begingroup$ Can you post a plot here? $\endgroup$ – endolith Feb 23 '15 at 17:47
1
$\begingroup$

This probably should be a comment, but I didn't earn that option yet. You can try measuring instrument impulse response. If I understand correctly, your system is essentially one-dimensional. Then if you use thin plate or plate of known thickness instead of thick object/material, your output signal will be an approximation of impulse response. Using it as deconvolution kernel will allow you to measure response specific to material, or object.

$\endgroup$
0
$\begingroup$

I would suggest looking into Adaptive Signal Processing techniques like RLS. http://researchtrend.net/ijet21/ijetnew/9%20Binod%20K.pdf

Component separation with ICA is possible provided the signal sources have the following characteristics.

  1. They are non-gaussian.
  2. Statistically independent from each other.
  3. Time of Arrival are approximately the same.
  4. Number of signal sources is at-least as large of number of estimated components.(You will have to account for the noise in the echo signal)

The drawbacks are the correct ordering of the estimated components, scaling and sign cannot be determined with ICA.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.