Modeling of PEMFC and Analysis of Multiple Influencing Factors on Output Characteristics

Abstract

The output characteristics of the Proton-Exchange Membrane Fuel Cells (PEMFCs) are affected by multiple factors, but quantitatively describing the relationships is challenging. In this paper, a semi-empirical dynamic model of PEMFC is established firstly. The influence of a single factor on the output characteristics of PEMFC is analyzed longitudinally. Then, a derivative significance weight analysis based on support vector regression (SVR-DSWA) algorithm is proposed to analyze the influence weights of multi-factors on the output characteristics, and the optimal parameters combinations in different current density regions are obtained by maximizing the output voltage values based on formulated SVR model. The Root-Mean-Square Error (RMSE) of output voltage prediction results based on the SVR algorithm is less than 0.0458, and the accuracy of weight analysis results by using the SVR-DSWA algorithm and the optimal parameters combinations analysis method are verified by 4-factor 3-level orthogonal experiments in low, medium, and high current density regions. The SVR-DSWA algorithm and optimal parameters combinations analysis method can replace the orthogonal experiment to analyze the influence weights and optimal combinations of input factors on the output characteristics within the full current density range rapidly, and has much higher efficiency than the orthogonal experiment. The analysis results can provide theoretical support for improving fuel cell performance and formulating a control strategy.