| Abstrakt: |
The detection of underground water leaks that deteriorate pipes is crucial for water authorities worldwide. Two factors are of utmost importance, i.e. the cost and the time required to locate the link. Numerous techniques have been proposed for detecting underground water leaks, with acoustic-based methods showing the most promise, particularly those using cross-correlation. However, cross-correlation relies on convolution, which necessitates extensive computation and large memory space. This paper addresses those two limitations by implementing a modified version of cross-correlation. In this modified version, the two signals to be correlated are first converted into a square shape with only two levels, either one or zero. When both signals are in agreement (either both ones or both zeros), they are added together, and when they are in disagreement (one is one and the other one is zero), they are subtracted. In leak detection, the strength of the peak is not as important as the location of the peak. By employing the current technique, the peak location obtained is the same as that obtained from classical correlation. [ABSTRACT FROM AUTHOR] |
