The monitoring of nuclear safeguards measurements consists on verifying the coherence between the operator declarations and the corresponding inspector measurements on the same nuclear items. Significant deviations may be present in the data, as consequence of problems with the operator and/or inspector measurement systems. But they could also be the result of data falsification. In both cases, quantitative analysis and statistical outcomes may be negatively affected by their presence, unless robust statistical methods are used. This article aims to investigate the benefits deriving from the introduction of robust procedures in nuclear safeguards context. In particular, we will introduce a robust estimator for the estimation of the uncertainty components of the measurement error model. The analysis will prove the capacity of robust procedures to limit the bias in simulated and empirical contexts, and consequently to provide more sounding statistical outcomes. For these reasons, the introduction of robust procedures may represent a step forward in the still ongoing development of reliable Uncertainty Quantification methods for error variance estimation.

CERASA Andrea; 

2022-08-31

MDPI

JRC129290

1099-4300 (online),   

https://www.mdpi.com/1099-4300/24/8/1160,    https://publications.jrc.ec.europa.eu/repository/handle/JRC129290,   

10.3390/e24081160 (online),