Remaining Useful Life Prediction of Linear Discrete Systems using an Interval Observer

This paper presents a new prognosis approach to estimate the Remaining Useful Life (RUL) of systems subject to output noise and disturbance. This approach comprises two phases: in the diagnosis phase, the fault is detected and estimated, and the prediction phase is used to predict the evolution of this fault in the future. The diagnosis step is carried out using an interval observer designed to estimate an addition to the complete state of a sensor fault on a for discrete linear systems. Next, based on an appropriate identification technique, a parametric identification is performed to estimate the future degradation until it reaches a certain failure threshold. This allows us to calculate the interval including the RUL. A numerical example illustrates the effectiveness and validity of the proposed technique