A novel prognostics system is developed based on a data-model-fusion scheme.
Bayesian inference-based probability is developed as a health indication of machine.
High-order particle filtering integrating logistic regression predicts machine health.
The results illustrate the potential applications of the proposed prognostics system.