Risk Assessment for Calibration: The Non-Linearized Case

Dubravka Božić, Biserka Runje, Andrej Razumić

Abstract

This study presents a novel mathematical model designed to evaluate risks in calibration processes. A probabilistic model was described for estimating the global producers' and consumers' risk with a non-linearized tolerance interval. The measurement uncertainty required to assess the risk for the explanatory variable at a given value of the response variable was calculated from the regression analysis outcome. The model was tested on the example from industrial practice to evaluate the risks associated with calibrating the roundness measurement device within a moderate scale of -3 μm to 3 μm. The results indicate that global consumer and producer risks are lower in the non-linear model than in the linear model. The calibration risk assessment model was evaluated using confusion matrix-based metrics.

Download

IMEKO-TC8-11-24-2025-007.pdf

DOI

10.21014/tc8-2025.007

IMEKO TC

TC8 - Traceability in Metrology

Back to the proceedings