A Reliable Fault Diagnosis Method Based on Belief Rule Base Consider Robustness Analysis

Researchers propose a new fault diagnosis method using belief rule base (BRB) technology with enhanced robustness analysis to improve the reliability of equipment monitoring systems. The approach addresses sensor uncertainty and model vulnerability, demonstrating improved accuracy and robustness in real-world applications like diesel engine and bearing fault detection.

This research addresses a critical gap in industrial equipment maintenance by tackling the robustness challenges inherent in sensor-based fault diagnosis systems. Equipment failures can cascade into costly production shutdowns, making reliable early detection essential for operational continuity. Traditional fault diagnosis methods struggle when sensor data becomes unreliable or operating conditions deviate from training scenarios, a vulnerability the proposed belief rule base approach systematically addresses through three optimization constraint strategies.

The technical contribution centers on quantifying and improving the robustness of diagnostic models rather than merely optimizing for accuracy. This distinction matters significantly because a highly accurate model trained on ideal sensor conditions often fails catastrophically when real-world noise, sensor drift, or unexpected operating scenarios emerge. By incorporating robustness analysis as a primary objective alongside accuracy, the method creates more reliable diagnostic systems that maintain performance across varying conditions.

For industrial operators and maintenance teams, this advancement translates to reduced false positives in fault alerts and improved detection of genuine equipment degradation. The validation using WD615 diesel engines and bearing datasets demonstrates applicability across different equipment types and failure modes. Organizations relying on predictive maintenance programs could benefit from more trustworthy diagnostic systems that reduce unnecessary interventions while catching actual faults earlier. This work contributes to the broader industrial automation and Industry 4.0 ecosystem by making sensor-based diagnostics more dependable. Future developments might focus on scaling these methods to more complex industrial systems and integrating them with real-time decision-making platforms.

• →Belief rule base method incorporates systematic robustness analysis to address sensor reliability challenges in fault diagnosis.
• →Three robustness constraint strategies improve both accuracy and resilience of diagnostic models across varying operating conditions.
• →Validated approach demonstrates effectiveness on real industrial equipment including diesel engines and bearing systems.
• →Enhanced robustness reduces false positives while improving detection of genuine equipment failures.
• →Applicable to predictive maintenance programs seeking more trustworthy sensor-based diagnostic systems.