Dynamic neural network-based system identification of a pressurized water reactor

This work presents a dynamic neural network-based (DNN) system identification approach for a pressurized water nuclear reactor. The presented empirical modelling approach describes the DNN structure using differential equations. Local optimization algorithms based on unconstrained Quasi-Newton and interior point approaches are used in the identification process. The efficacy of the proposed approach has been demonstrated by identifying a nuclear reactor core coupled with thermal-hydraulics. DNNs are employed to train the structure and validate it using the nuclear reactor data. The simulation results show that the neural network identified model is sufficiently able to capture the dynamics of the nuclear reactor and it is suitably able to approximate the complex nuclear reactor system.