Online joint estimation and prediction for system-level prognostics under component interactions and mission profile...

Area of Science:

• Engineering
• Reliability Engineering
• System Dynamics

Background:

• Predictive maintenance relies on accurate failure prognostics for effective asset management.
• Current prognostics often focus on individual components, neglecting system-level interactions and mission profile impacts.
• System-level prognostics remain an underexplored area crucial for comprehensive maintenance strategies.

Purpose of the Study:

• To develop an online joint estimation and prediction methodology for system-level remaining useful life (RUL).
• To address the limitations of component-level prognostics by incorporating component interactions and mission profile effects.
• To provide a framework for real-time RUL estimation with minimal prior knowledge.

Main Methods:

• Utilized the inoperability input-output model (IIM) framework to capture system dynamics and component interdependencies.
• Implemented an online estimation process using a gradient descent algorithm for real-time parameter updates.
• Recursively updated system parameters with new measurements to refine RUL predictions.

Main Results:

• Demonstrated the effectiveness of the proposed online joint estimation and prediction methodology through numerical examples.
• Successfully applied the approach to the Tennessee Eastman Process, validating its performance in a real industrial application.
• The method accurately predicts system RUL by accounting for component interactions and mission profile effects.

Conclusions:

• The proposed IIM-based methodology offers a robust solution for system-level failure prognostics.
• Online estimation and prediction enhance the accuracy and applicability of RUL determination in complex systems.
• This approach advances predictive maintenance by enabling more informed operational and maintenance decisions.