Intelligent agents for the optimal operation of water distribution pumping stations

Water distribution systems have become increasingly complex and more difficult to operate. With ever increasing regulations and pressure over efficient quality service, the automation of control centers seems to be the objective of the next decades for water utilities. The field of Artificial Intelligence has been exploited to develop online decision support systems that interact directly with pressure and flow sensors as well as with actuators such as valves and pumps. In this thesis, a number of control scenarios are presented, where intelligent agents could be used to further optimize control objectives and provide the operators with reliable tools to make a more efficient use of the available resources. One of these scenarios was selected to be studied further: the real-time operation of pumping stations. Prototype simulators and control agents were implemented in order to assess the potential benefits that these technologies promise. An incremental level of difficulty was used for the agent to first balance between energy and maintenance costs, then to detect and recover from failures, and finally to be able to cope with stochastic demands like those in real systems. As opposed to offline implementations, online agents must deal with a number of additional complications. However, the results show that Artificial Intelligence has indeed a great potential for water distribution automation, welcoming future developments to address this still adolescent research line.

• Autores: Hernández Cruz, Felipe
• Fecha de publicación: 2010
• Institución: Universidad de los Andes
• Repositorio: Séneca: repositorio Uniandes
• Idioma: eng
• OAI Identifier: oai:repositorio.uniandes.edu.co:1992/11302
• Acceso en línea: http://hdl.handle.net/1992/11302
• Palabra clave: Inteligencia artificial - Investigaciones; Control en tiempo real - Investigaciones; Estaciones de bombeo - Modelos de simulación - Investigaciones; Ingeniería
• Rights: openAccess
• License: http://creativecommons.org/licenses/by-nc-nd/4.0/