Abstract
Model-assisted probability of detection (MAPOD) is the key metric for reliability analysis of nondestructive testing (NDT) systems. Fast metamodeling techniques advances the MAPOD process by efficiently capturing the physics and then largely reducing the number of physics-based model evaluations. This work presents the MAPOD analysis through the polynomial chaos-based Kriging (PCKriging) metamodel. In particular, the proposed PCKriging approach is demonstrated on an analytical function and an ultrasonic testing benchmark case and compared against the current state-of-the-art metamodels. Preliminary results in this work show that the PCKriging is capable of reducing the training cost by two to four times fewer than the current state of the art.
How to Cite:
Du, X., Leifsson, L. ., Nagawkar, J., Meeker, W., Gurrala, P., Song, J. & Roberts, R., (2019) “Metamodel-based uncertainty propagation for model-assisted probability of detection”, Review of Progress in Quantitative Nondestructive Evaluation .
Downloads:
Download PDF