固有直交分解による電気化学ポテンシャル解析の代理モデルの構築

抄録

In this paper, we study the prediction of behavior using proper orthogonal decomposition (POD) to oxygen potential in a solid oxide fuel cell (SOFC). Transient electro-chemical potential analyses of the SOFC with some fuel mass inflow and temperature as input parameters are performed and oxygen potentials that changes over time are obtained. POD is applied to the data matrix of oxygen potentials and a reduced model that extracts the dominant mode is created. Then, by using function approximation for the reduced model, a surrogate model is developed to predict the efficient behavior of the oxygen potential in SOFC under different conditions. To validate the surrogate model, we compare the result of surrogate model with the one of electro-chemical potential analysis under arbitrary condition, in which hydrogen pressure and temperature are distributed in SOFC under this condition.