Acoustic Water Leak Detection System

Liselene de Abreu Borges, Miguel Arjona Ramírez

DOI: 10.14209/sbrt.2010.21

Evento: VII International Telecommunications Symposium (ITS2010)

Keywords: Water Leak Detection SVM AR PCA

Abstract

This article presents an acoustic water leak detection system. It shows all the stages for doing acoustic leak signal processing and classification. The technique used for classification is support vector machine (SVM) using as features the Itakura mean and maximum distance and the larger projection of the principal component analysis (PCA). The whole process is based on the autoregressive (AR) models of the signals. Detection rate achieved is 80%.

Download