Sliding mode observer based incipient sensor fault detection with application to high-speed railway traction device

Area of Science:

• Control Systems Engineering
• Fault Detection and Diagnosis
• Nonlinear System Analysis

Background:

• Incipient sensor faults pose significant risks in complex nonlinear systems.
• Observer-based fault detection is challenged by unmatched uncertainties.
• Ensuring system stability and robustness is critical for reliable operation.

Purpose of the Study:

• To develop an effective incipient sensor fault detection scheme for nonlinear systems with observer unmatched uncertainties.
• To design a robust sliding mode observer that maintains sliding motion despite faults.
• To enhance fault detectability using novel adaptive thresholds.

Main Methods:

• Design of a sliding mode observer for an augmented system including incipient sensor faults.
• Parameter design using Linear Matrix Inequalities (LMI) and line filter techniques.
• Proposal of three levels of adaptive thresholds based on reduced-order sliding mode dynamics.

Main Results:

• Generated residuals are robust to system uncertainties and faults.
• Sliding motion is preserved, ensuring observer functionality.
• Significantly improved detectability of incipient sensor faults.

Conclusions:

• The proposed fault detection scheme effectively addresses incipient sensor faults in nonlinear systems.
• The method demonstrates robustness against uncertainties and maintains system stability.
• Successful application to a China Railway High-speed traction system validates the approach.