par Herrera, Manuel ;Ramos-Martínez, Eva;Gutiérrez-Pérez, Joanna J.A.;Izquierdo, Joaquín;Pérez-García, Rafael
Référence Mathematical Modeling in Social Sciences and Engineering, Nova Science Publishers, Inc., page (81-90)
Publication Publié, 2014-04
Partie d'ouvrage collectif
Résumé : Nowadays biofilm develops in all drinking water distribution systems (DWDSs) and leads to various undesirable problems, representing a paradigm in management of these systems. Biofilm formation is dependent on a complex interaction of various factors, those associated with the DWDSs infrastructure being the most flexible and adaptable to minimize biofilm development in the inner pipe walls. One of the main objectives in the quality control of DWDSs is the analysis of biofilm development in the network, discovering areas more prone to it. This chapter proposes spectral clustering to achieve these issues after a kernel abstraction of the available data. The theoretical approach is enriched by both: an entropy based ranking to select the eigenvectors to work with, and a full tuning process that automatizes the proposal. Finally, a real case-study is introduced to check the performance of this division of the network. The results are promising, showing the relative trend to biofilm development for each sector. Using this information, managers can carry out, if neccesary, changes in the DWDS infrastructure to minimize biofilm development and perform biofilm control policies more efficiently.