Real-Time Residual Life Prediction Based on Adaptive Wiener Process with UKF

Life span prediction is of significant importance for enhancing the reliability and safety during the use of equipment. In this study, we introduce an innovative degradation model, grounded in the Wiener process, for accurately estimating the Remaining Useful Life (RUL) of systems, offering enhanced predictive capabilities compared to existing approaches. Initially, the variability of degradation drift is considered to construct a state-space model for online updating of the equipment's remaining life. Subsequently, an analytical expression for the remaining life in terms of the first passage time is derived, and the Expectation-Maximization (EM) method is applied for estimating the model parameters. Finally, an example of fatigue crack growth demonstrates that the constructed model has a good accuracy in estimating the RUL.