Improving nuclear safeguards evaluation through enhanced belief rule-based inference methodology

The present work aims to propose a framework for modelling, analysing and synthesising information on nuclear safeguards with various uncertainties. After a hierarchical analysis of states' nuclear activities on the basis of the International Atomic Energy Agency (IAEA) Physical Model, the multilayer structure of the evaluation model for states' nuclear activities is outlined. Special emphasis is given to the synthesis and evaluation of the Physical Model indicator information by using a newly developed belief rule- based inference methodology (RIMER), which handles hybrid uncertain information in nuclear safeguards' evaluation process. RIMER is developed on the basis of the Dempster-Shafer theory of evidence, decision theory and fuzzy rule-based systems. In this methodology, various types of information from different sources can be transformed and used in the inference process. The new methodology can provide a flexible and effective procedure to deal with hybrid, uncertain assessment information to arrive at rational conclusions. The evaluation model framework illustrates and clarifies the inference and synthesis formalism from the given case study in this paper.