Expert Opinion Fusion Framework Using Subjective Logic for Fault Diagnosis

Area of Science:

• Engineering
• Artificial Intelligence
• Uncertainty Quantification

Background:

• Fault diagnosis is crucial for engineered systems maintenance.
• Traditional Bayesian networks (BNs) struggle with second-order uncertainty.
• Existing models may yield unreliable diagnosis results when uncertainty is high.

Purpose of the Study:

• To propose a hierarchical system diagnosis fusion framework.
• To explicitly address and manage uncertainty using subjective logic (SL).
• To improve the reliability of fault diagnosis in complex systems.

Main Methods:

• Designed individual subjective Bayesian networks (SBNs) for expert knowledge representation.
• Clustered experts into groups based on knowledge similarity.
• Employed a one-opinion fusion method to combine SBN inferences for consensus diagnosis.

Main Results:

• The proposed fusion framework demonstrated superior performance compared to state-of-the-art methods.
• Achieved stable diagnostic performance across various simulated scenarios.
• The framework effectively handles uncertainty, leading to more reliable fault diagnosis.

Conclusions:

• The subjective logic-based fusion framework offers a robust solution for fault diagnosis.
• It advances the state-of-the-art in diagnosing complex engineered systems.
• This approach provides a promising direction for future research in uncertainty management.