Fault prognosis is a key enabler of predictive maintenance in modern industrial systems, where heterogeneous sensing, modeling, and data analytics coexist under varying operating conditions. This paper proposes a reliability-aware health index fusion framework for hybrid fault prognosis that systematically integrates physics-based, signal-based, data-driven, and statistical prognostic methods within a unified probabilistic formulation. Each prognostic output is mapped to a bounded health index, while method-specific, stage-dependent reliability is learned offline from run-to-failure data using confusion matrices over discretized health states. During online operation, health state estimates are fused using a Bayesian time-recursive framework that accounts for degradation dynamics and reliability variation. Simulation-based case studies on rotating machinery demonstrate that the proposed approach significantly improves health index estimation accuracy and reduces variance compared to individual prognostic methods, particularly near failure.
Azizi et al. (Thu,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: