Fault detection and location

Context - Issues: Modern methods for detecting and localizing failures (sensors, actuators, physical components) based on models, analytical redundancy, monitoring/diagnostics

Objectives: Analyze the tasks of a monitoring system. Master projection methods in parity space and based on diagnostic observers. Monitoring-oriented modeling.


Tools: lectures, paper-based tutorials, Matlab/Simulink mini-project.

Basics of condition assessment and observation.

* Introduction to model-based monitoring systems
- FDIR systems: Fault Detection, Isolation and Recovery.
- Model-based FDI, methodological approach:
Generation of fault-indicating signals
Decision-making, hypothesis testing
Robustness / Performance.
- Examples and fields of application:
Aeronautical systems
Space applications
Nuclear sector
Mechatronic systems
Environmental sector.
- Detectability ("signal" point of view, "system" point of view).

* FDI by projection of redundancy relations in parity space
- Static redundancy
- Dynamic redundancy
- Decoupling, "Gantmacher" optimization procedure.

* FDI based on diagnostic observers
- Dedicated Observer Scheme
- Generalized Observer Scheme.

* Decision testing, indicator signal processing
- Notion of risk and hypothesis testing
- Sequential testing (real-time), Wald test.

Automatic

Powerpoint slides.

Recommended books: Gertler J. (1998). "Fault detection and diagnosis in engineering systems" Marcel Dekker.
Chen, J. and R.J. Patton (1999). Robust model-based fault diagnosis for dynamic systems. Kluwer Academic Publishers.