Quality Classification of Green Pellet Nuclear Fuels using Radial Basis Function Neural Networks

Total quality classification process is necessary to be continously conducted along the pellet fabrication processes to minimize the number of rejected of the green pellets. This cylindrical uranium dioxide pellets, as the main fuel element in the Light Water Nuclear Reactor, should shows uniform shape, uniform quality and a high density profile. The quality of green pellets is conventionally monitored through a laboratory measurement of the physical pellets characteristics followed by a graphical chart classification technique, however, this technique is difficult to use and shows low accuracy and time consuming. In this paper, a Radial Basis Function neural networks is develop by studied and modified the weight initialization on its neural structure, and applied for automation of classifying the pellets quality. It is proved from the experiments that the Radial Basis Function neural networks shows a comparable classification rate with that of best-tune Back Propagation neural networks, however, the computational cost is reduced significantly.