Optimization of Two-Dimensional Porosity Distribution of Cathode Gdl for Performance Enhancement of Pemfcs Using the Sine Function

Abstract

This paper introduces trigonometric functions  as the equation for porosity distribution. By controlling the values of A, B, C, and D, a series of unique one-dimensional (1D) porosity distribution equations for the cathode GDL along the FC-CL and cathode inlet directions are generated. Along different directions, it is divided into ten equal-length intervals, in each interval the average porosity is calculated, and it is used as a new feature value introduced into the support vector regression (SVR) algorithm for learning. The model performs well, with an R2 of 0.99. Using a 1D SVR model, the optimal 1D porosity distribution of the cathode GDL is predicted, and the simulation results show an error of less than 0.1%. Further, the 1D porosity distributions along the FC-CL and cathode inlet directions are combined to form a two-dimensional (2D) porosity distribution, which is divided into 20 different new feature values through the same operation, and the weight of the average porosity is increased, ultimately the R2 value of the SVR model is 0.97. The optimized 2D SVR model is used to predict and output the corresponding optimal porosity distribution. And the 2D optimal porosity distribution is compared with the 1D distribution, deriving the 2D optimal porosity distribution equation and cloud map of GDL at the FC-CL and cathode inlet of the single-channel PEMFCs. By comparing the optimal porosity distribution with the uniform porosity distribution, it is found that the proportion of current density and oxygen concentration increase significantly, while the growth of temperature and water content is smaller, indicating that the optimal porosity distribution enhances the performance of PEMFCs.

Keywords: Trigonometric functions, porosity distribution equation, SVR algorithm, optimal porosity distribution.