Tek-Tips is the largest IT community on the Internet today!

Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!

  • Congratulations MintJulep on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!

cross-correlation function - vibration

Status
Not open for further replies.

lodo123

Civil/Environmental
Joined
Sep 22, 2005
Messages
1
Location
CA
Hi,

I'm trying to determine the source of a vibration signal by the use of 2 accelerometers, a dual channel spectrum analyzer and the cross-correlation function. The length of the time signal is 10s and my frequency range is 5000 Hz.

As the Cross correlation function is a measure of the similarity in two time domain signals, I am expecting a result close to 1 if the signals are identical and if they are completely dissimilar, the cross correlation will be zero.

My first step was to verify this method. I put the 2 accelerometers close to one source (motor) and the results with the cross-correlation function was, roughly, a line starting starting from 0.02 to 0.

Did I miss something or is my asumption wrong?

On the other hand, the coherence (frequency domain) has the results that I expected: value of 1 except at the resonant frequencies on the tested piece of equipment (motor).

I tried different configuration and set-up but I still have the same results with the cross-correlation...

Thanks for your help,

Laurent.

 
not quite sure how your software works. But according to the cross-correlation equation. The cross-correlation also depends on the time delay \tao. For your case, it is 1 only if the time delay between the two signal is zero. There might be a hidden default parameter set wrong in your software.
 
Use coherence in the frequency domain, not cross correlation in the time domain.

We use partial coherence for source identification - measure the vibration on each wheelhub, and the noise in the car. The partial cohernece then tells you which frequency comes from which wheel. If you are lucky.

However it doesn't work as well with periodic signals from multiple sources.







Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.
 
Try to correlate x1(t) to x2(t-1),x2(t-2),...
M777182
 
Are you sure you are looking at cross-correlation or are you looking at cross-covariance? x-correlation is scaled from 0 to 1 x-covariance is not.

M

--
Dr Michael F Platten
 
Cross-correlation should give you the relative position of the source with respect to the two receivers (accelerometers) assuming you know the speed of sound in the medium or structure being tested.
The result of a cross-correlation analysis is time, the difference between the time required for the vibration to reach accel 1 versus the time to reach accel 2, from the source position. Knowing the time gives the difference in distance from source to the two accelerometers. Knowing this and distance between the two accelerometers allows solving for the distance from each accelerometer to the source.
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top