The use of random-fuzzy variables for the implementation of decision rules in the presence of measurement uncertainty

The practical, everyday final applications of measurement processes are mostly aimed at taking a decision, on the basis of a comparison between the measured value and a reference value. If uncertainty in measurement is considered, this comparison must be performed between an interval of confidence (the measurement result) and a scalar quantity (the reference value). The result of such a comparison is quite often not univocal, so that taking a decision may become quite troublesome. This paper shows how the use of the random-fuzzy variables in the expression of uncertainty in measurement allows to implement simple decision rules capable of taking into account the measurement uncertainty correctly. The proposed decision rules are applied to measurement procedures based on measurement algorithms that contain if ... then ... else structures where the if condition is applied to intermediate measurement results. An example of implementation of these decision rules is reported and discussed.