A technique based on uncertainty analysis to qualify the design of measurement systems

In the paper the authors propose an original approach for uncertainty estimation in decision-making problems. By an innovative analysis of incorrect decision probabilities associated with the considered process, the used Statistical Model allows to estimate the maximum uncertainty that can affect the measurement system. Starting from statistical information on the process, this procedure provides a practical tool for qualifying the measurement system design in order to assure desired performances and a suitable confidence level for decision-making stage. In particular, during the synthesis phase of measurement system, the model permits to settle the allowed system's metrological characteristics, according to required incorrect decision probabilities. The general hypotheses formulated on the model make it applicable in process control field to a wide variety of measurement systems, when a generic quality index is put in comparison with a reference limit. The power of the used approach is the opportunity of estimating the allowed measurement system uncertainty to reduce the impact of the incorrect decision probabilities on decision-making stage. Moreover the employment of fuzzy logic makes possible to model the vague features associated with the process to be monitored. © 2005 IEEE.