From diagnosis to maintenance and prognosis

Prognosis is of great interest for industrial applications and especially for preventive maintenance. Diagnosis and prognosis are closely related: diagnosis determines the set of faulty configurations that are consistent with a given set of observations whereas prognosis determines a faulty configuration that will be reached in the future and that is consistent with the actual diagnosis and with the ageing model of the supervised system. These techniques aim at determining the health state from which the remaining useful life can be evaluated. They may rely on physics-based models, statistical data analyses, learning and/or estimation methods.

A lot of work has to be done in this field in order to better understand and benefit of the relationships between diagnosis and prognosis: logical characterization, prognosis methods relying on reliability concepts, definition of prognosability, etc.