Prognosis of Uncertain Discrete-Time Nonlinear Systems for Remaining Useful Life Estimation
摘要
This paper deals with a new prognosis methodology based on an interval multi-observer of a nonlinear system subject to disturbances and output noise. The RUL of the system in a degraded situation is then determined using the interval approach. The idea is based on two steps: In the first, a diagnostic algorithm is designed to detect and estimate the fault in addition to the full system state. This estimation is realized over a past receding window where a polynomial equation is used to approximate the sensor fault behavior. In the second step, the system dynamic is predicted over a moving prediction horizon using the system model and the approximate fault model identified in the first step to obtain the lower and upper bounds of the remaining useful life. A numerical example is presented to show the efficiency, performance, and accuracy of the proposed methodologies.