Modeling Superconductive Fault Current Limiter Using Constructive Neural Networks

Although so many advances have been proposed in the field of artificial intelligence and superconductivity, there are few reports on their combination. On the other hand, because of the nonlinear and multivariable characteristics of the superconductive elements and capabilities of neural networks in this field, it seems useful to apply the neural networks to model and control the superconductive phenomena or devices. In this paper, a new constructive neural network (CNN) trained by two different optimization algorithms; back-propagation and genetic algorithm, is proposed which models the behavior of the superconductive fault current limiters (SFCLs). Simulation results show that the proposed approach is in good harmony with the real characteristics of the SFCLs.