ARMA Modeling and Bootstrap Technique for Accuracy Evaluation of Parameters of Interest in Nuclear Reactor Field

摘要

In Nuclear Reactor Engineering field, ARMA modeling of recorded signals has come into wide use and represents a well established technique for estimating parameters of interest. In lack of information about the probability distribution of the recorded data, the parameter's accuracy may be evaluated by the recently proposed Bootstrap method, suitably worked out, when only a single short record of data is available. The present paper deals with such a procedure, working on the white residuals of the recorded series. The method is checked at first against non-normality by operating on simulated series with exponential white noise. The procedure is then applied to thermocouple experimental data measured in a sodium cooled subassembly mock-up of a fast reactor. The aim is to evaluate the accuracy of the skewness coefficient, whose potential as indicator of modification in the radial temperature profile has been recognized. (ERA citation 12:009114)