A new safety analysis approach using enhanced mathematical sturctures

摘要

In probabilistic safety analyeses, the input data and logical combinations of the data can involve considerable subjectivity. We have begun an investigation of some innovative mathematical structures that are ptoentially useful for incorporating subjectivity into safety analyses and have developed some practical illustrations of the utility of the structures. A particular advantage of the approach is that these results are combined with conventional mathematical structures in a hybrid analysis. This enables us to incorporate subjectivity along with known data and physical models and to present the output in a form that directly indicates the portion of subjectivity in the final outcome. Some of the mathematical structures investigated are based on various logical norms, Ordered Weighted Averaging, implication operations, Gamma operators, sigmoid operators, inference networks, failure "race" comparatorsm min/max ordinate logic, threshold logic, conditional possibility, and Frechet-based dependence operators.