Fuzzy clustering application on failure rate prediction in water distribution networks

In this report a new approach of failure rate prediction is presented based on Fuzzy Clustering technic for a more deterministic and accurate implementation of neuro-fuzzy systems. This technique is compared with two benchmark methods: Artificial Neural Networks (ANN) and Adaptative Neuro-Fuzzy Infe...

Descripción completa

Detalles Bibliográficos
Autores: Vendrell Gallart, Joan, Sun, Congcong, Puig Cayuela, Vicenç|||0000-0002-6364-6429, Cembrano Gennari, Gabriela|||0000-0003-1436-6022
Tipo de recurso: informe técnico
Fecha de publicación:2020
País:España
Institución:Universitat Politècnica de Catalunya (UPC)
Repositorio:UPCommons. Portal del coneixement obert de la UPC
Idioma:inglés
OAI Identifier:oai:upcommons.upc.edu:2117/330935
Acceso en línea:https://hdl.handle.net/2117/330935
Access Level:acceso abierto
Palabra clave:Control theory
Water networks
Prediction
Control
Classificació INSPEC::Control theory
Àrees temàtiques de la UPC::Informàtica::Automàtica i control
Descripción
Sumario:In this report a new approach of failure rate prediction is presented based on Fuzzy Clustering technic for a more deterministic and accurate implementation of neuro-fuzzy systems. This technique is compared with two benchmark methods: Artificial Neural Networks (ANN) and Adaptative Neuro-Fuzzy Inference Systems (ANFIS). Furthermore, an analysis of the necessary inputs is carried out with the goal of defining the useful information needed for the models. All these methods are applied to real data of Barcelona water distribution system and the models predictions are compared with calculated pipe failure rate.