Smart Tuning of Predictive Controllers for Drinking Water Networked Systems
This thesis affords the tuning of a multi-objective predictive controller, particularly designed for the Barcelona’s drinking water network. Predictive controller objectives have been established taking into account the minimisation of three criteria; the first one considers economic costs involved...
| Autor: | |
|---|---|
| Tipo de recurso: | tesis de maestría |
| Fecha de publicación: | 2010 |
| 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:2099.1/16319 |
| Acceso en línea: | https://hdl.handle.net/2099.1/16319 |
| Access Level: | acceso abierto |
| Palabra clave: | Drinking water Water-supply – Automatic control Predictive control Aigua potable –- Abastament –- Control automàtic Control predictiu Àrees temàtiques de la UPC::Desenvolupament humà i sostenible::Enginyeria ambiental::Tractament de l'aigua Àrees temàtiques de la UPC::Informàtica::Automàtica i control |
| Sumario: | This thesis affords the tuning of a multi-objective predictive controller, particularly designed for the Barcelona’s drinking water network. Predictive controller objectives have been established taking into account the minimisation of three criteria; the first one considers economic costs involved in the distribution process; second one takes into account tank’s safety volumes; and the third penalises excessive variations in control actions. Disturbances usually affect the operational conditions in which an automatic control strategy evolves along time. In the drinking water network (DWN) predictive control strategy, consumer demands have been modelled as measured disturbances. It is primordial to ensure an effective rejection of those measured disturbances, respecting to the controller objectives. Through this thesis, functionalities of the DWN system are illustrated as well as the predictive control strategy used to solve the multi-objective optimisation problem; later, methods to explore the space of non-dominated solutions, known as Pareto front, are exposed and a strategy to choose, at every sample time, a solution in line with the problem objectives. Next, a tuning strategy, which avoid the Pareto front calculation in the on-line implementation by using a model that allows to variate the weighting factors of each objective function, in terms of consumer water demands. Keywords: Model predictive control, large-scale systems, drinking water networks, multiobjective optimisation. |
|---|