Removing model and data non-conformity in measurement evaluation
暂无分享,去创建一个
A procedure is derived to remove after adjustment a frequently observed non-conformity of the evaluation model and the given information. This procedure is nearly as easy to apply as the commonly used Birge ratio procedure but it has the advantage of being founded on the Bayesian theory of measurement uncertainty. The derivation of the procedure rests on the shown feature of the generalized least-squares analysis of measurement data to be the only adjustment method in measurement evaluation satisfying the metrological requirements of unique solution and consistent uncertainty analysis in accordance with recent international recommendations. Information conservation is a particular case of consistency. This feature and the procedure follow from the inherent symmetry of the suitably linearized evaluation model and of the information given by the measurement data and the associated uncertainties.
[1] B. Taylor,et al. CODATA Recommended Values of the Fundamental Physical Constants 1998*$\dagger$ , 1999 .
[2] K. Weise,et al. A Bayesian theory of measurement uncertainty , 1993 .
[3] Peter J. Mohr,et al. CODATA Recommended Values of the Fundamental Constants: 1998 , 2000 .
[4] W. Michaelis,et al. A new precise determination of Newton's gravitational constant , 1995 .
[5] Werner A. Stahel,et al. Robust alternatives to least squares , 1996 .