Keywords: horizontal steam generators, recurrent neural networks, simulation, VVER nuclear reactors, nuclear power plants, nuclear energy, thermohydraulics, nuclear safety, safety analysis, safety systems, control systems, steam generator behaviour
Identification of a VVER horizontal steam generator using recurrent neural networks
Although a large number of simulations have been developed for U-tube steam generators in Western pressurised water reactors, no suitable models yet exist for horizontal steam generators in VVER-type nuclear power plants. This research aims to use RELAP5 thermohydraulic code data to identify a horizontal steam generator via an intelligent structure named Nonlinear Auto Regression with eXogenous inputs (NARX). Simulation results show that this method can predict the steam generator behaviour faster than RELAP5 with an acceptable error level. The identified model can be used in future research in safety analysis, safety systems and control systems for predicting steam generator behaviour.