Life Prediction Method of Nonlinear Degradation Equipment Based on Adaptive Unscented Particle Filter

To solve the problem that degradation process is changeable, a life prediction method based on adaptive unscented particle filter (AUPF) is proposed. Firstly, the nonlinear stochastic degradation process based on the state space model is established. According to the concept of first arrival time and the fault probability density function (PDF), the approximate life distribution is given. Then, a real-time degradation state and noise estimation algorithm based on unscented particle filter (UPF) and Sage-Husa filter is proposed to update the life distribution online. Finally, the validity of the prediction method is verified by simulation experiments and life prediction of motorized spindle.