Improved Sensor Fault Detection, Isolation, and Mitigation Using Multiple Observers Approach.

Area of Science:

• Control Systems Engineering
• Fault Diagnosis
• System Monitoring

Background:

• Traditional Fault Detection and Isolation (FDI) relies on steady-state residual signals, limiting its effectiveness.
• Existing methods struggle with transient states and faults in critical, unobservable sensors.
• System safety is compromised if faulty sensor mitigation is delayed.

Purpose of the Study:

• To propose advanced FDI methods addressing limitations of traditional residual-based techniques.
• To enable fault detection during observer transient states and for critical sensors.
• To integrate new methods with existing approaches for a robust FDI and mitigation framework.

Main Methods:

• Analyzing estimation error convergence rates for transient state fault detection.
• Employing open-loop observers to detect faults in critical sensors.
• Implementing observer switching strategies for fault impact mitigation.

Main Results:

• Developed three novel methods to enhance FDI capabilities.
• Integrated these methods with a residual-based approach, creating an improved framework.
• Validated the framework mathematically and through simulations on a 5-state suspension system.

Conclusions:

• The proposed integrated FDI framework effectively detects and isolates sensor faults, including those in critical sensors and during transient states.
• The novel methods enhance system safety by enabling earlier fault detection and mitigation.
• The approach offers a significant improvement over traditional residual-based FDI techniques.