This paper introduces the problem of probability dilution in satellite conjunction analysis to the wider uncertainty quantification community and proposes random-to-fuzzy transformation as a practical fix. Probability dilution is presented as an "ostrich paradox," a situation in which poor data quality appears to make a system safer, an absurd conclusion. Plausibility of collision derived from a judiciously chosen random-to-fuzzy transformation provides a more intuitive metric for the possibility of a collision between two satellites. Two candidate transformations are presented, provisionally labeled "red fuzzy" and "blue fuzzy." One of these transformations (blue) emulates the current maximum probability approach. The other transformation (red) yields clear indications as to when the user is in the probability dilution region. The paper concludes with a case in favor of the red random-to-fuzzy transformation as the blueprint for a usable full-scale solution.
展开▼
