A Fuzzy Approach to represent Measurement Data affected by both A-type and B-type Uncertainty

This paper proposes a new approach to transform the probability distribution of N repeated measurement results into a possibility distribution in presence of both A-type and B-type uncertainty. The technique presents the final measurement result in terms of a fuzzy set which easily expresses the levels of confidence for the final uncertainty to be determined as a-cuts. In this treatment the statistical independence of the measurement data permit the assumption that the samples are derived from a normal distribution. The fuzzy set representation of measurement uncertainty is preferable in cases where acquisitions have to be further processed and where, in these events, a conversion into symbolic format may be also necessary reasoning systems. As a case study, the propagation of uncertainty in presence of decision rules is considered