 |
Efficient Estimation of A-basis and B-basis under Limited Data and Simplified Models
Estimation of A/B-basis values is essential for composite structure certification, traditionally grounded in extensive experimental datasets. However, this work extends this concept to numerical models, addressing the specific challenges of mixed aleatory and epistemic uncertainties inherent in virtual testing. We specifically target three sources of epistemic uncertainty : model identification, statistical estimation, and surrogate error. The proposed framework ensures an unbiased, efficient estimator and enables robust confidence bounds.
