A Likelihood-Based Approach of Uncertainty Quantification Using Both Sparse Point Data and Interval Estimates

In this paper, we propose a likelihood-based approach to quantify the uncertainty in the presence of mixed uncertainties, which can be applied in varieties of PHM cases such as reliability evaluation and system fault prognosis. By modeling intervals as additional evidence, non-probability information is incorporated into a unified Bayesian model. The proposed method outperforms existing studies both in accuracy and uncertainty reduction. This approach could benefit system fault prognosis, warranty policy making as well as maintenance service planning, etc. A numerical case is demonstrated for illustration and validation purposes.