Modelling of solid oxide electrolyser cell using extreme learning machine

•The steady-state performance of SOEC is tested under different gas compositions.•The SOEC performance is modelled by extreme learning machine algorithm.•The analyses of SOEC performance and modelling results are provided.

Solid Oxide Electrolyzer Cell (SOEC) can covert H2O and/or CO2 into usable fuel by consuming the excess electricity of renewable resource or off-peak grid power. The SOEC is a promising device for the sustainable development of energy and hydrogen economy. In this work, the steady-state performance of SOEC is tested under different gas compositions and modelled by extreme learning machine (ELM) algorithm. According to the experimental results, the concentrations of H2O and CO2 influence the performance of SOEC. For the model, the inputs are the operating voltage and volume percentage of H2, CO2, and H2O, while the output is the performance (current) of SOEC. The obtained model has correlation coefficients of higher than 0.999 and root mean square error less than 0.018, which means that the predicted data by the model well matches the experimental results. Then, the obtained ELM model is used to analyse the performances of SOEC under different concentrations of feedstock. Thus, this data driven ELM model is suitable for many instances of fast modelling for individual group and may be helpful to save the cost, time and effort to build a model for the purpose of performance analysis and system level design.