Recursive zonotopic set-membership approach for system-level prognostics with application to linear parameter-varying...

Area of Science:

• Engineering
• Control Systems
• Reliability Engineering

Background:

• System degradation necessitates accurate Remaining Useful Life (RUL) forecasting for proactive maintenance.
• Existing methods often struggle with nonlinearities and unknown uncertainties in degraded systems.
• Formulating prognostics as a system-level problem is crucial for complex machinery.

Purpose of the Study:

• To propose a robust recursive zonotopic set-membership approach for RUL forecasting.
• To address system-level prognostics for components exhibiting degradation.
• To handle nonlinear systems reformulated as linear parameter-varying (LPV) models.

Main Methods:

• Augmenting critical degraded components to system states, creating a nonlinear system.
• Reformulating the nonlinear system into a linear parameter-varying (LPV) model.
• Employing joint state and parameter estimation within a zonotopic set-membership framework.
• Utilizing optimal linear matrix inequality (LMI)-based tuning for the estimation scheme.
• Assuming unknown-but-bounded (UBB) noises and uncertainties.

Main Results:

• A recursive zonotopic set-membership approach for RUL forecasting was successfully developed.
• The methodology accurately predicts failure precursors in degraded systems.
• Validation on a DC-DC converter case study demonstrated high estimation and forecasting accuracy, even with unknown degradation behaviors.

Conclusions:

• The proposed recursive zonotopic set-membership approach offers a robust solution for RUL forecasting in LPV systems with degraded components.
• The method effectively handles system nonlinearities and uncertainties through joint state-parameter estimation.
• The successful application to a DC-DC converter highlights the practical applicability and accuracy of the prognostics methodology.