Tutorial: Data-driven diagnostics and prognostics

Fault diagnosis and prognosis (FDP) plays an important role in the modern complex industrial systems. Diagnosis aims to monitor the fault state in real-time while prognosis predicts the evolution of fault state and remaining useful life (RUL). Traditional Riemann sampling-based FDP (RS-FDP) takes samples and executes algorithms periodically and, in most cases, requires significant computational resources, which makes it difficult to be implemented on hardware with very limited computational capabilities. To overcome this bottleneck, a Lebesgue sampling-based FDP (LS-FDP), in which FDP algorithms are implemented “as-needed”. In LS-FDP, a set of Lebesgue states are defined on the state axis. The computation of LS-based diagnosis is triggered only when the value of measurements changes from one Lebesgue state to another, or “event-triggered#x201D;. This method significantly reduces the computation demands by eliminating unnecessary computation. This LS-FDP design is generic and able to accommodate different FDP algorithms. In this presentation, the design of LS-FDP and its application to engineering systems will be discussed in details. The efficiency of LS-FDP is verified by comparison with those of its RS-FDP counterparts.