Abstract
In this thesis we propose a method for diagnosis problem solving. This method is based on a multi-modeling approach describing both normal and abnormal behavior of a system. This modeling approach allows to represent a system at different abstraction levels (behavioral, functional and teleological). Fundamental knowledge is described according to a bond-graph representation. We show that bond-graph representation can be exploited in order to generate (completely or partially) the functional models. The different models of the multi-modeling approach allows to define the functional state of a system at different abstraction levels. We exploit this property to exonerate sub-systems for which the expected behavior is observed. The behavioral and functional descriptions of the remaining sub-systems are exploited hierarchically in a two steps process. In a first step, the abnormal behaviors explaining some observations are identified. In a second step, the remaining unexplained observations are used to generate conflict sets and thus the consistency based diagnoses. The modeling method and the diagnosis process have been applied to a Reactor Coolant Pump Sets. This application illustrates the concepts described in this thesis and shows its potentialities. (authors)
Citation Formats
Thetiot, R.
A diagnosis method for physical systems using a multi-modeling approach; Utilisation de l'approche multi-modeles pour l'aide au diagnostic d'installations industrielles.
France: N. p.,
2000.
Web.
Thetiot, R.
A diagnosis method for physical systems using a multi-modeling approach; Utilisation de l'approche multi-modeles pour l'aide au diagnostic d'installations industrielles.
France.
Thetiot, R.
2000.
"A diagnosis method for physical systems using a multi-modeling approach; Utilisation de l'approche multi-modeles pour l'aide au diagnostic d'installations industrielles."
France.
@misc{etde_20131241,
title = {A diagnosis method for physical systems using a multi-modeling approach; Utilisation de l'approche multi-modeles pour l'aide au diagnostic d'installations industrielles}
author = {Thetiot, R}
abstractNote = {In this thesis we propose a method for diagnosis problem solving. This method is based on a multi-modeling approach describing both normal and abnormal behavior of a system. This modeling approach allows to represent a system at different abstraction levels (behavioral, functional and teleological). Fundamental knowledge is described according to a bond-graph representation. We show that bond-graph representation can be exploited in order to generate (completely or partially) the functional models. The different models of the multi-modeling approach allows to define the functional state of a system at different abstraction levels. We exploit this property to exonerate sub-systems for which the expected behavior is observed. The behavioral and functional descriptions of the remaining sub-systems are exploited hierarchically in a two steps process. In a first step, the abnormal behaviors explaining some observations are identified. In a second step, the remaining unexplained observations are used to generate conflict sets and thus the consistency based diagnoses. The modeling method and the diagnosis process have been applied to a Reactor Coolant Pump Sets. This application illustrates the concepts described in this thesis and shows its potentialities. (authors)}
place = {France}
year = {2000}
month = {Jul}
}
title = {A diagnosis method for physical systems using a multi-modeling approach; Utilisation de l'approche multi-modeles pour l'aide au diagnostic d'installations industrielles}
author = {Thetiot, R}
abstractNote = {In this thesis we propose a method for diagnosis problem solving. This method is based on a multi-modeling approach describing both normal and abnormal behavior of a system. This modeling approach allows to represent a system at different abstraction levels (behavioral, functional and teleological). Fundamental knowledge is described according to a bond-graph representation. We show that bond-graph representation can be exploited in order to generate (completely or partially) the functional models. The different models of the multi-modeling approach allows to define the functional state of a system at different abstraction levels. We exploit this property to exonerate sub-systems for which the expected behavior is observed. The behavioral and functional descriptions of the remaining sub-systems are exploited hierarchically in a two steps process. In a first step, the abnormal behaviors explaining some observations are identified. In a second step, the remaining unexplained observations are used to generate conflict sets and thus the consistency based diagnoses. The modeling method and the diagnosis process have been applied to a Reactor Coolant Pump Sets. This application illustrates the concepts described in this thesis and shows its potentialities. (authors)}
place = {France}
year = {2000}
month = {Jul}
}