6. Predictive maintenance decision methods

6.1 Principles

In predictive maintenance, decision theory enables managers to select thresholds reliably and not "by guesswork", by making a compromise between two major risks: false alarms, which lead to unjustified and costly equipment shutdowns, and non-detection, which corresponds to the continued operation of faulty equipment, with potentially critical consequences.

Decision-theoretic methods are based on the formulation of hypotheses (H ₀ : normal / H ₁ : failure) and the assessment of risks in terms of error probabilities (false alarm, non-detection). A large number of methods have been developed using statistical criteria and tests or Bayesian approaches, and are...