Research on Masked Multiple Fault Method Based on 5-SDG

Diagnosing multiple faults in complex industrial systems can be challenging. Multiple faults are categorized into induced faults, independent multiple faults, masking faults, and correlated multiple faults based on the causal relationships between fault sources. The diagnosis of masking faults can be particularly difficult due to their resemblance to single faults but with varying effects on neighboring nodes. The SDG fault diagnosis method is a prevalent approach used to diagnose multiple faults, which graphically depicts the complex interrelationships among system state variables. This paper proposes a 5-SDG-based masked multiple fault model, leveraging the correlations between different types of state variables in multiple fault scenarios to enhance the model's expressive power. Furthermore, a fault inference and judgment method is proposed based on this model in masked multiple fault scenarios. Experimental validation on a three-tank water tank model confirms the effectiveness of the proposed approach.