Skip to Main content Skip to Navigation
Theses

Diagnostic des systèmes industriels complexes par agrégation de méthodes : application à une station d'épuration

Abstract : This thesis proposes a hierarchical approach for the complex systems diagnosis based on the decomposition in elementary sub-systems of diagnostic and their aggregation. The chapter 1 gives details of the real-time diagnosis problems as those of the complex systems. It proposes a solution by hierarchical decomposition and presents the software tool DIALOGS stemming from this concept. It presents at last a complex system illustrating our approach, the waste water treatment plant of Maxéville and its sludge thermal drying process. The chapters 2 and 3 tackle the modular aspect of this approach through the symptoms generation step. The chapter 2 proposes for this sorne detection methods making independent symptoms. The chapter 3 presents complex techniques which allow faults isolation through symptoms group. The necessity to associate the residuals coming from these methods in higher level of decision imposes to homogenise them by the mean of decision tests, chapter 2. The chapter 4 develops the hierarchical notion and the aggregation of elementary systems through two faults isolation methods, the structured residuals and the multiple purposes decision. A third technique, the generic decision , allows to conclude as for the complex system state.
Document type :
Theses
File URL :
http://docnum.univ-lorraine.fr/prive/SCD_T_1998_0002_GIRAUD.pdf
Complete list of metadata

https://hal.univ-lorraine.fr/tel-01747042
Contributor : Thèses Ul <>
Submitted on : Thursday, March 29, 2018 - 10:49:58 AM
Last modification on : Tuesday, January 26, 2021 - 2:38:50 PM

Identifiers

  • HAL Id : tel-01747042, version 1

Collections

Citation

Denis Giraud. Diagnostic des systèmes industriels complexes par agrégation de méthodes : application à une station d'épuration. Autre. Université Henri Poincaré - Nancy 1, 1998. Français. ⟨NNT : 1998NAN10002⟩. ⟨tel-01747042⟩

Share

Metrics

Record views

24