Hybrid Stochastic Models for Remaining Lifetime Prognosis Dissertation

The dissertation focuses on creating analytical models to estimate the remaining lifetime probability distribution of components in fluctuating environments. It explores three stochastic process models: temporally nonhomogeneous Markov, temporally homogeneous Markov, and semi-Markov environments. By integrating real-time degradation data from sensors with these models, it computes lifetime distributions, revealing that certain models yield matrix-exponential type distributions. To address computational challenges in the semi-Markov case, phase-type approximations are employed. The findings demonstrate the potential of these techniques for lifetime prognosis across various applications.