The unscented transform: a powerful tool for measurement uncertainty evaluation

An original approach for uncertainty evaluation in indirect measurements is presented here in after. The approach applies the unscented transform to the measurement model (i.e. the functional relationship between the output and input quantities) in order to estimate the output expectation along with the associated standard uncertainty. Thanks to some nice properties of the unscented transform, notable limits of the current GUM recommendations are overcome. In particular, reliable estimates are granted also in the presence of nonlinear and/or non-analytical measurement models, or complex digital signal-processing algorithms. A number of numerical tests are conducted on simulated and actual measurement data. Remarkable concurrence between the obtained estimates and those granted by Monte Carlo simulations confirms the reliability and efficacy of the proposed approach