A Cognitive Framework for Analysis and Treatment of Uncertainty in Prognostics

Uncertainties exist in fault prognostics systems can lead to inaccurate results and this will lead to unnecessary or delay maintenance activities. The uncertainty must be considered carefully to achieve more effective engineering applications. Uncertainties have been classified as aleatory uncertainty and epistemic uncertainty. Aleatory uncertainty is also called objective uncertainty, irreducible uncertainty, inherent uncertainty, and stochastic uncertainty. Epistemic uncertainty is also referred to as subjective uncertainty, reducible uncertainty and state-of-knowledge uncertainty. A cognitive framework to aid in the understanding of uncertainties and techniques for mitigating and even taking positive advantage of them is presented. From the perspective of man-machine-environment system engineering, the framework is an attempt to clarify the wide range of uncertainties that affect prognostics system. The uncertainty sources are identified as three aspects (machine, environment, man). A general uncertainty management procedure is proposed. It mainly contains uncertainty identification, qualification, propagation and sensitivity analysis. For case illustration purpose, the popular data-driven prognostics methods are discussed in detail. Current and developing methods for dealing with uncertainties are projected onto the framework to understand their relative roles and interactions.