Diagnosability of distributed systems or how to improve maintenance activities in an aircraft. (pdf)

Nowadays, one economical challenge for leading aircraft manufacturers such as Airbus is about the improvement of the maintenance activities on any aircraft in terms of efficiency and costs. The longer the maintenance operations are, the more costly they are as the aircraft is not operational during the maintenance phase. The technical challenge is then to minimize the maintenance time by minimizing the number of maintenance operations on ground these operations should only be the replacement of faulty components and nothing else. To achieve this, it is thus required to embed in the aircraft a monitoring and diagnosis system that is in charge of determining on-board the faulty components to replace. The less ambiguous the diagnosis is, the more efficient the maintenance operations are.

Aircrafts are complex systems resulting from the assembling of heterogenous components designed by different companies. It follows that the development of a diagnosis architecture (sofware agents, communication protocols) over the whole aircraft is challenging as it requires to provide local design recommandations to any companies so that the whole system is as diagnosable as possible. This problem is the so-called sensor placement problem. In this talk, we present a theoretical framework that defines the sensor placement problem in a distrbuted discrete event system and propose a methodology that firstly provides design recommendations in order to guarantee diagnosability objectives and secondly assists in the design of the corresponding monitoring and diagnosis architecture.

This work is part of the ARCHISTIC project in collaboration with Pauline Ribot, Michel Combacau from LAAS and Airbus and Ecole Nationale d'Ingenieurs de Tarbes, France.