Skip to Main content Skip to Navigation
Theses

Identification de systèmes non linéaires par une approche multi-modèle : application à la modélisation de la relation pluie-débit pour le diagnostic de fonctionnement de capteur

Abstract : Mathematical models are often required for process control, behaviour prediction, fault diagnosis, estimation of unmeasurable variables and improving understanding of system behaviour. Many real processes involve many variables, are non-linear and vary over time. Moreover, for such systems, knowledge of the underlying physical phenomena is often incomplete and modelling has to be performed from observed date. Variousmodelling and identification techniques based on fuzzy logic and neural networkds have been developed for non-linear dynamic systems. In this thesis, the multi-model approach has been investigated. Different aspects of parameter estimation and structure identification of a quasi-linear multi-model are addressed. The proposed approach has been applied for modelling the rainfall-runoff relationship on urban watersheds located in Nancy. Actual rainfall and runoff data have been extracted from the database of metropolitan authority sewer control centre. The obtained models have then been implemented into fault detection and isolation procedures of sensors involved in the sewer network management, namely, tipping-bucket rain gauges and water level gauges
Document type :
Theses
File URL :
http://docnum.univ-lorraine.fr/prive/INPL_T_1998_BOUKHRIS_A.pdf
Complete list of metadata

https://hal.univ-lorraine.fr/tel-01752246
Contributor : Thèses Ul <>
Submitted on : Thursday, March 29, 2018 - 1:39:39 PM
Last modification on : Wednesday, February 3, 2021 - 1:18:06 PM

Identifiers

  • HAL Id : tel-01752246, version 1

Collections

Citation

Anass Boukhris. Identification de systèmes non linéaires par une approche multi-modèle : application à la modélisation de la relation pluie-débit pour le diagnostic de fonctionnement de capteur. Autre. Institut National Polytechnique de Lorraine, 1998. Français. ⟨NNT : 1998INPL060N⟩. ⟨tel-01752246⟩

Share

Metrics

Record views

23