Power supply quality improvement with a SOFC plant by neural-network-based control

This paper demonstrates the potential of a solid-oxide fuel cell (SOFC) to perform functions other than the supply of real power to the grid. These additional functions however require the use of an inverter. The flux-vector control is used very effectively for the control of this inverter, where the space-vector pulsewidth modulation (SVM) is implemented by neural networks (NNs). The results presented in the paper show the effect of the fuel cell on the voltage at the sensitive load point. The performance of the fuel cell was found to be excellent.