Diagnosis and health monitoring

Lecturer(s): Emmanuel BOUTLEUX, Catherine MUSY, Olivier ONDEL
Course ⋅ 16 hStudy ⋅ 12 h


To detect failure before they appear is a big challenge for any kind of complex systems. From modern car full of automation (sensors, actuators, control/command strategies) to more-electric airplanes, from industrial power plant to robotics applications, methods are needed to inform that a failure or default as appeared, appears or will appear. That course will focus on automatic detection methods based on model-based approaches or artificial intelligence approaches.


Diagnosis, health monitoring, identification, pattern recognition,FMECA


Context Fonctional approches like FMECA (Failure Modes, Effects and Criticality Analysis) Reliability Diagnosis approaches:

  • model-based
    • identification
    • error detection -artificial intelligence
    • pattern recognition
    • clustering
    • decision rules Perspectives

Learning Outcomes

  • To realise challenges and difficulties associated with health monitoring
  • To be able to applied pattern recognition techniques
  • To be able to properly identify mathematical model for diagnosis purposes
  • To be able to select parameters identification methods


Final mark = 50% Knowledge + 50% Know-how Knowledge = final exam Know-how = average mark issued from 3 reports from BE