Fundamentals of neural networks: architectures, algorithms, and applications
Fundamentals of neural networks: architectures, algorithms, and applications
Fuzzy Sets and Systems - Special issue on nuclear engineering
Hi-index | 0.00 |
This paper discusses about the implementation of an Artificial Neural Network (ANN) based model for Neutronics power estimation of Prototype Fast Breeder Reactor (PFBR). ANN has been designed to predict the Neutronics power for various positions of control and safety rods. Simulation studies were carried out using the thermo-hydraulics code and the required data has been generated for training the ANN model. Here, three layer neural network architecture has been developed and trained with different learning algorithms to estimate Neutronics power and model the plant dynamics. The best performing algorithm in terms of faster convergence has been identified among the variants of back propagation network. ANN model has more advantages compared to conventional model namely increased speed, reduced complexity in addition to producing accurate results. Neural network being a data driven model, it is possible to derive the operating characteristics of nuclear reactor without exploring the intricacies of the complex subsystems.