抄録
In this paper, we use the Baysian approach to evaluate the uncertainty of a measurement based on quantization data. We assume that the quantity before quantization obeys a normal distribution having the average μ and standard deviation σ. Based on a posterior probability estimated by n-times repetitive measurement, we are able to determine the probability density p (μ, σ) using the Baysian method. An uncertainty of measurement corresponds to a probability function concerning parameter μ. The probability function has been used to determine the estimation value after quantization and its uncertainty.
This paper describes the computational results for events such that the number of repeated measurement is smaller than fifteen and no more than three quantized values occur in each event. When all of the measurement data take the same value, the results show the conventional Type B evaluation described above results in underestimation of the uncertainty if the number of data is equal to four or less.
Furthermore, we show that estimations performed using this approach are significantly different from those performed by Type A evaluation when the event has indicating values for two ranged values.
