A Methodology for Discerning Incipient Boiling of the Coolant in a Water-Moderated, Water-Cooled (Pressurized-Water) Nuclear Reactor by Means of the Bayesian Neutron-Noise Classifier

Noise signals at the output of the technological parameters of a nuclear power plant (neutron flux, dynamic pressure, etc.) contain important information on the stale of the equipment. Effective algorithms have been developed for identifying random processes which, after appropriate transformation, are treated as multidimensional random vectors. These vectors are automatically classified by means of the aforementioned algorithms on the basis of the likelihood function, in particular, of the Bayesian classifier and resolving functions. The use of the Bayesian classifier is based on constructing multidimensional distributions of probabilities in the attribute space with the applicable dimensionality, which serves to describe random representing vectors of the applicable images that are subject to automatic classification.