Remaining Useful Life Prediction Method of Aero Engine With Multilayer Uncertainty

Uncertainties associated with the prediction of the Remaining Useful Life (RUL) of random degradation equipment are influenced by such factors as time-varying uncertainty, individual difference, and measurement error. Given this, a predictive method for the RUL of an aero -engine with three layers of uncertainty was proposed. Firstly, historical condition monitoring data was used to generate a Composite Health Index (CHI) for characterizing the performance degradation of the engine. Then a nonlinear Wiener degradation model is built considering three layers of uncertainty. Secondly, the maximum likelihood method is applied to obtain the estimates of the priori distribution of the random coefficients in the degradation model. Then, the degradation states were updated synchronously by applying the Kalman Filtering (KF) algorithm and constructing the state-space model. Finally, the Probability Density Function (PDF) of the RUL with three layers of uncertainty was deduced from the total probability formula. A numerical example and a case study comparing several representative methods in the literature were presented using the aero-engine data. The simulation example analysis shows that the proposed method can significantly improve RUL prediction accuracy, and thus it has a particular engineering application value.