Abstract
A revised method is proposed for processing measurements on the basis of a generalized normal distribution for the random error: a three-parameter distribution very close to the sampling distribution for the measurements that at the same time preserves all the advantages of a gaussian distribution. The method has been tested out on various state and secondary standards. It produces a substantial effect when one needs to obtain a measurement result with the maximum possible accuracy.
Similar content being viewed by others
References
Guide on the Expression of Measurement Uncertainties [Russian translation], GUP VNIIM im. D. I. Mendeleeva, St. Petersburg (1999).
I. G. Fridlender, “The distribution of random measurement errors,” in: Proceedings of Zaporozhe Agricultural Engineering Institute, Issue 2 [in Russian], Mashgiz, Kiev (1955), p. 13.
A. É. Fridman, Izmer. Tekh., No. 11, 3 (1991).
H. Kramer, Mathematical Methods of Statistics [Russian translation], Mir, Moscow (1975).
GOST 8.207-76, The State System of Measurements: Direct Measurements with Repeated Observations: Methods of Processing the Observational Results: Basic Concepts [in Russian].
Rights and permissions
About this article
Cite this article
Fridman, A.É. A New Methodology for Processing Repeated Measurements. Measurement Techniques 44, 1149–1157 (2001). https://doi.org/10.1023/A:1014025720672
Issue Date:
DOI: https://doi.org/10.1023/A:1014025720672