CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Recognizing a random signal (http://www.cfd-online.com/Forums/fluent/46981-recognizing-random-signal.html)

cfd_newbie January 9, 2008 22:15

Recognizing a random signal
 
How may I decide if a signal is random or not ? The usual way is to plot the signal versus time and decide if it looks random.

There is a mathematical definition for a random signal ? If I want to analyze numerical a signal how may I decide if it is random ?

Thanks in advance.

anders January 10, 2008 08:55

Re: Recognizing a random signal
 
This has nothing to do with cfd, but anyway... You should preform a fast fourier transform on the signal and then plot it (preferably with Matlab or similar software). If it is a random signal no frequence will differ while a non-random one will have one or several peaks (one peak for each present frequency).


cfd_newbie January 10, 2008 12:46

Re: Recognizing a random signal
 
Thanks Anders, actually my problem has connection with CFD because I try to analyze the results from a DNS calculation. I need a criteria for identifying for example if the pressure at a certain point is turbulent or not.

dave January 10, 2008 18:26

Re: Recognizing a random signal
 
Typically a random signal has a very narrow autocorrelation peak (in either space or time) and the fourier components have no obvious phase relationship to one another. You can also do chi-squared tests to evaluate whether the signal obeys a particular statistical distribution (log-normal, gaussian, poisson, etc). Finally you can analyse whether the statistical moments (mean, variance, etc. ) converge as the number of data points increases.



All times are GMT -4. The time now is 12:33.