Evaluation of measurement uncertainty based on Bayesian information fusion

This paper raises a new method for evaluating uncertainty that taking count of both the record and the data. By using Bayesian Statistical Principle, the prior distribution and the posterior one, provided by the record and the data, were combined together. The statistical characteristics parameter estimation was descended from the posterior distribution, so that a formula of the uncertainty, which combined the advantages of type A and B, was acquired. By simulation and verification, this measurement shows great advantages compared with the others, especially to small size of data analysis.