Fault Diagnosis of Nuclear Power Plant Based on Genetic-RBF Neural Network

This paper presents development of an automatic fault diagnosis system in the nuclear power plants to minimize possible nuclear disasters caused by inaccurate diagnoses done by operators. Combined binary and decimal coding methods are employed in this work based on radial basis function neural network (RBFNN) structure. This underling RBFNN structure is further trained through genetic optimization algorithm based on known frequent failure conditions from a nuclear power plant's condensation and feed water system. It is found that the proposed Genetic-RBFNN (GRBFNN) method not only makes the original neural network smaller in terms of computation and realization but also improves diagnosis speed and accuracy.