User's Manual

68 I 72
In general, however, the respective signatures arrive at the sensors at different times and
simultaneously recorded signals do not correlate with one another. Because the signals were
digitally recorded and stored, it is possible to retroactively shift the signals with respect to one
another and re-correlate them with one another.
This is performed systematically. From the results, one obtains a correlation curve that
includes the individual correlation values over the respective computationally performed time
shift.
If, during this process, one reaches precisely the time shift that corresponds to the
propagation time difference from the sound source to the two receivers, the signal signatures
match and one obtains a correlation maximum at this point.
Because this is the comparison of two different signals, it is also called a cross correlation.
In the context of leakage noises, the goal of the correlation measurement is primarily to
determine the propagation time difference of two signals.
Because the correlation curve also includes spectral information, it is often of interest to
define the character of the leakage noise, particularly for distinguishing noises that are not
related to the leak, provided this can be detected by means of a frequency analysis (e.g.,
electrical noise). The correlation curve can also be mathematically filtered for the purpose of
retroactively separating such interfering noise to better identify the correlation maximum
caused by the leak.