Title
Identifying household water use through transient signal classification Identifying household water use through transient signal classification
Author
Faculty/Department
Faculty of Applied Engineering Sciences
Publication type
article
Publication
New York ,
Subject
Engineering sciences. Technology
Source (journal)
Journal of computing in civil engineering. - New York
Volume/pages
(2015) , p. 04015007-1-04015007-10
ISSN
0887-3801
Carrier
E
Target language
English (eng)
Full text (Publishers DOI)
Abstract
The research reported in this paper aims to develop a household water use identification method through signal pattern analysis. An experimental facility was constructed to simulate bathroom and kitchen water use. The data acquisition system used a volumetric water meter with pulsed output, pressure transducers, data acquisition with a Universal Serial Bus interface interconnected with the Cyble sensor and a laptop computer. The data analysis was performed using a pattern recognition algorithm to identify the hydraulic fixtures in use. Five classes of water use were considered, as follows: (1) kitchen faucet (KF), (2) washbasin faucet (WF), (3) bidet (BD), (4) shower (SH), and (5) toilet flush (TF). Two algorithms were used to identify the best classifier for the data, as follows: (1) multilayer perceptron, and (2) support vector machine (SVM). The fusion by majority vote regarding the results of SVM in the time domain showed the best accuracy; 92% accuracy for kitchen faucet, 94% for washbasin faucet, 94% for bidet, 100% for the shower, and 100% for toilet flush, thus supporting the use of signal signatures of flow and pressure in identifying the hydraulic fixtures in use.
E-info
https://repository.uantwerpen.be/docman/iruaauth/aad88d/128666.pdf