A prognostic framework based on factorial Hidden Markov Model

In this paper, a system level fault diagnosis and prognosis framework is proposed. Test delay and prognostic dependency relationship are considered in a model based on Factorial Hidden Markov Model. By using time change, the inference problem can be transformed into a known one and thus can be solved using the previous algorithms. The proposed model and algorithm are tested on a simulated model, which shows that they have the potential to be applied to fault diagnosis and prognosis.