Skip to Main content Skip to Navigation

Méthodologie de surveillance dynamique à l'aide des réseaux neuro-flous temporels.

Abstract : Our work concerns the industrial monitoring, process usually parse into two phases : the detection and the diagnosis. We thus propose a dynamic monitoring aid system based on two tools of Artificial Intelligence technics. The first one used for the dynamic detection is a recurrent radial basis function network. The second one, based on a neuro-fuzzy network, perform the diagnosis aid.
With the sensor data, the detection too determine the degree of possibility of each operating mode of the system. Given these information, the diagnosis tool searchs causes by an abductive approach, and classify them by a degree of credibility. For the configuration and the initialization of the tools, we used some historic and maintenance data of the system like AMDEC and Fault Tree.
The development part of this thesis is divided into two points : a software integration on an industrial computer (UML approach + implementation) and the application on a flexible production system.
Document type :
Complete list of metadatas

Cited literature [101 references]  Display  Hide  Download
Contributor : Martine Azema <>
Submitted on : Friday, January 25, 2008 - 11:07:21 AM
Last modification on : Thursday, November 12, 2020 - 9:42:34 AM
Long-term archiving on: : Thursday, April 29, 2010 - 6:19:55 PM


  • HAL Id : tel-00217474, version 1



Nicolas Palluat. Méthodologie de surveillance dynamique à l'aide des réseaux neuro-flous temporels.. Automatique / Robotique. Université de Franche-Comté, 2006. Français. ⟨tel-00217474⟩



Record views


Files downloads