Electric Performance Model of Solid Oxide Electrolytic Cell Based on Neural Network

Solid oxide electrolysis cell (SOEC) has become a hot research topic as an efficient hydrogen production scheme. When the system works in variable working conditions, its electrolytic voltage will change with the current density, but the electrolytic voltage will be affected by temperature, gas partial pressure and other parameters, resulting in inconsistent I-V relationship of the system under different working conditions. In this paper, a neural network model of 3 input–1 output is established to predict the electrolytic voltage required by the system under different operating parameters under different operating conditions through a small amount of system simulation data. The results show that the prediction error is less than 1%, and the neural network method is suitable for the prediction model of electrolytic voltage demand in SOEC system. The model can be used for dynamic control analysis of SOEC system under variable conditions.