Lifetime Prediction Based on Age-State Dependent Diffusion Model

As a key component of prediction and health management, remaining useful life prediction is critical to improving system reliability, availability and safety. Accurate life prediction is based on a reliable prediction model. Most of the existing studies are based on age-dependent models. However, the degradation process of a system is not only influenced by age but also by the system state in real situations. To address this problem, an age-state dependent diffusion model is proposed in this paper. The maximum likelihood estimation algorithm is used to perform parameter estimation. The validity of the proposed model is verified by simulation data set.