Performance improvement of solid oxide fuel cell by neural network and multi-objective optimization algorithm

Channel structure design usually plays an important role in improving the performance of solid oxide fuel cell (SOFC). A new channel structure of SOFC with spherical obstacles was proposed in this work. First, compared with the rectangular obstacle channel, the power density of SOFC with spherical obstacles increased by 0.86%, the pressure drop decreased by 31.1%, and the temperature gradient decreased by 4.29%, which means the SOFC with spherical obstacles has better performance. Then, the effect of the configuration parameters, operational parameters, and porosity on the performance of SOFC was studied, and the satisfactory surrogate model with a maximum error smaller than 5% was obtained by back propagation neural network and Latin hypercube sampling to predict the performance of SOFC with spherical obstacles. Finally, the multi-objective optimization technique was employed by non-dominated sorting genetic algorithm and the linear programming method for multidimensional analysis of preference method to improve the performance of SOFC with spherical obstacles. The results showed that the SOFC with optimized spherical obstacle channel obtained a maximum power density of 2809.9 A/m2 and a minimum temperature gradient of 5.89 K/cm, improving the power density by about 6.7% and decreasing the temperature gradient by 12.09% in comparison with the original spherical obstacle channel. Overall, the present work provides an optimized channel design approach to improve the performance of the SOFC.