A model based on bootstrapped neural networks for computing the maximum fuel cladding temperature in an Rmbk-1500 nuclear reactor accident

摘要

The present paper illustrates the use of artificial neural networks for computing the maximum fuel cladding temperature reached during a complete group distribution blockage scenario in an RBMK-1500 nuclear reactor. The uncertainties associated to the neural predictions are quantified by resorting to the bootstrap technique. The trained neural networks are further used to perform a sensitivity analysis aimed at identifying the parameters which most significantly influence the maximum fuel cladding temperature.