Brief description of past research projects

Coordination: Nicolas Guelfi and Pierre Plumer Chef of Project: Marcos Da Silveira Partners: University of Luxembourg, Public Research Center Henri Tudor. This project contributes to improve the design, analysis and implementation of healthcare web services. The aims of RESIST are: The identification of the needs of medical sector in Luxembourg in order to implement a home monitoring service; The selection of a set of technologies that can be used to implement this service; The definition of an architectural framework to support web-services analysis and design in order to improve the interoperability between medical applications; The definition and implementation of a case study.

Coordination: Louise Travé-Massuyès. Partners: LAAS-CNRS (Laboratory for Analysis and Architecture of Systems – French National Centre of Research), ALCATEL SPACE, Onera (Aeronautics and Space Research Center of Toulouse), CNES (French National Space Research Centre). Abstract: Embedded systems are achieving unprecedented levels of robustness by dramatically increasing their use of computation. Many of these systems will need to perform robustly within extremely harsh and uncertain environments, or may need to operate for years with minimal human interference. To achieve this, they need to have on-board fault detection and diagnosis (FDD) capabilities. Some traditional approaches are efficient for a limited set of systems. In this project, we aim to evaluate four different FDD techniques (Stochastic approach, Bounded incertitude approach, Qualitative Approach and Particle Filter based approach) and chose one or a mix of them to compose the fault detection, isolation and recovering system of an interplanetary satellite.

Coordination: Michel Combacau. Partners: LAAS-CNRS, HISPANO-SUIZA (SNECMA). Abstract: The objective of this project is to investigate how to diagnose complex dynamical system whose behavior is modeled as a hybrid system (discrete and continuous models). The method adopted consists in (1) identifying the hybrid system’s current state, (2) estimating the next probable state of the systems, (3) detecting values divergence and (4) identifying faults occurrences. In the specific system application (aircraft turbine) we used double redundant sensor in order to identify system and/or sensor faults. A set of common critical faults was introduced on the system to measure the reliability of proposed algorithm under different conditions and for different kinds of components. During the tests, gradual and abrupt faults were applied over temperature, rotation and pressures component and over on/off valves. Electrical and magnetic faults were introduced in one of the sensors. The results of these experiments were compared with a quality reference value and with values obtained from the method currently adopted by the partner. The performance of the proposed method was, in all cases, superior of the given reference and in most of the cases, it was superior of the method currently adopted.

The objective of this work is to propose a method to distribute centralized models described with Petri Nets in order to obtain a set of sub-models and relations that represents (each) part of the whole system. The obtained results also include the definition of a communication protocol that uses the remained redundancies of the model to keep the consistency of data and facilitate the supervision of the process. The performance of the method was evaluated according to a set of quality indicators and compared with the performance of the centralized system.

Coordinator: Marcos Da Silveira (From March 2004 until July 2004). Partner: Pontifical Catholic University. Abstract: The objective of this project was to model the behavior of a distributed factory manufacturing system (FMS), to define the execution and interaction rules, and to develop a recovery procedure (based on a multilevel supervisory system) for diagnosed failures. Petri nets and UML was adopted for modeling the plant behavior and the communication protocol. During the first semester of 2004, the physical implementation had been performed. One master student and two undergraduate students participated in this implementation. The plant was composed of 3 PLC (S-700), 12 pneumatic actuators (4 actuators controlled by each PLC), 2 electrical actuators and a large set of sensors. All PLC is connected with a PC (the supervisor) via a ProfBus FMS. I continued to participate in this project as a collaborator until July 2005.

Coordination: Marco Busetti de Paula Collaboration: Eduardo Loures and Marcos Da Silveira. Partner: FESTO Brazil. Abstract: This project had started with my master thesis in 1997 and had continued during the master studies of E. Loures in 1999. From there, two more master theses had improved this project. Its initial objective was to design and to implement a CACSD (Computer-Aided Control System Design) system in order to apply to Control Theory Education. The modularity and the friendship environment of this system attracted an industrial partner. New tools was proposed and implemented to introduce the “hardware-in-the-loop” method in the simulation module. This system uses a VXI hardware to offer a virtual instrumentation capacity and it interacts with MatLab/Simulink for data analysis and control simulations.