APPROCHE INTELLIGENTE À BASE DE RAISONNEMENT À PARTIR DE CAS POUR LE DIAGNOSTIC EN LIGNE DES SYSTÈMES AUTOMATISÉS DE PRODUCTION

Abstract : Automated Production Systems (APS) represent an important class of industrial systems. They have become complex and susceptible to malfunctions, with significant negative consequences on productivity, production quality and security of property and people. The main challenge when using such systems is to implement systemic assistance or diagnosis approaches in order to ensure their operating safety. Within this framework, we focus, in this thesis, on the online diagnosis of the APS equipped with sensors and actuators emitting binary signals. These systems can be considered as ‘Discrete Event Systems’ (DES). The effectiveness of an approach for diagnosing such systems is measured through good detection rate, isolation accuracy, false alarms number and the method implementation complexity. The objective of this work is, thus, to propose intelligent diagnosis solutions satisfying the mentioned criteria without complete knowledge about the system’s internal functioning. The solutions, whose implementation is not expensive, are able to update themselves in real time in order to improve their performance. The introduced approach is based on a reasoning and learning methodology derived from ‘Artificial Intelligence’ (AI) which is the ‘Case Based Reasoning’ (CBR). The originality of our research work is observed in the following 4 aspects : (1) the developed approach uses the CBR for the diagnosis of the APS with DES dynamics, (2) it suggests a case representation format inspired by the ‘Causal Temporal Signatures’ (CTS) which is able to adapt to the dynamic aspect of the systems to be monitored, (3) it exploits data acquired from a digital twin after its emulation in both normal and faulty modes to build an empirical knowledge called ’cases’ and finally (4) it presents a reasoning and learning phase that allows not only the online diagnosis of the monitored system, but also the updating of the case base following the appearance of the new unknown behaviors.
Complete list of metadatas

Cited literature [233 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/tel-02273482
Contributor : Nourhène Ben Rabah <>
Submitted on : Thursday, August 29, 2019 - 12:14:01 AM
Last modification on : Friday, August 30, 2019 - 1:17:09 AM

File

memoire.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : tel-02273482, version 1

Collections

Citation

N. Ben Rabah. APPROCHE INTELLIGENTE À BASE DE RAISONNEMENT À PARTIR DE CAS POUR LE DIAGNOSTIC EN LIGNE DES SYSTÈMES AUTOMATISÉS DE PRODUCTION. Intelligence artificielle [cs.AI]. UNIVERSITÉ DE REIMS CHAMPAGNE-ARDENNE et UNIVERSITÉ DE LA MANOUBA, 2018. Français. ⟨tel-02273482⟩

Share

Metrics

Record views

149

Files downloads

33