Improvement of Paramater Estimation Through Uncertainty Analysis

Whenever multiple independent estimates of a complex performance parameter are available, the question of which estimate to use or how to best combine the data invariably arises. This paper describes a methodology which utilizes the uncertainty intervals for each of the independent estimates to improve the overall estimate of the parameter and to minimize the uncertainty of the final answer. In addition, a simple validity test is described which can help identify instrumentation or other associated errors. Sensitivity analyses are included to show the effect of incorrect uncertainty (or error) analyses of the independent estimators. Monte Carlo results are presented in support of the capability of this methodology.