Parameter optimization of membrane mlectrode assembly in Fuel Cell based on improved differential evolution algorithm: biomechanical stress and strain considerations

The membrane electrode assembly in a proton exchange membrane fuel cell (PEMFC) functions as the electrochemical reaction region, where the generated electric current relies on the diffusion of reactant gases and electron conduction. Drawing inspiration from biomechanics, this study embarked on constructing a database of PEMFC performance data. Similar to how biomechanical studies use advanced imaging and sensing techniques to map the internal workings of organisms, three-dimensional computational fluid dynamics (CFD) simulations were employed to capture the intricate fluid and gas behaviors within the fuel cell. The data was then used to train data-driven surrogate models based on artificial neural network (ANN) and improved differential evolution for rapid prediction and optimization. When considering the biomechanical aspects, we analyze the mechanical stresses and strains that occur within the membrane electrode assembly during operation. These biomechanical factors can affect the durability and performance of the fuel cell. The gas diffusion layer (GDL) is similar to the pore structure in biological tissues. The pore structure of biological organisms, such as bones, not only ensures the diffusion and transport of nutrients, but also provides space for the attachment of cells to maintain the growth and metabolism of bones. The optimization results revealed that the pores of the GDL, just like the pores of biological tissues, affect the diffusion efficiency of the reactant gases (similar to nutrients) to the catalytic layer, and an appropriate porosity ensures the supply of the reactants required for the electrochemical reactions inside the cell, and improves the PEMFC performance of the cell. By utilizing the random forest algorithm (RF) to conduct feature importance evaluation, we can gain further understanding and interpretation of the factors influencing coupling relationships. The researchers successfully identified the optimal values of GDL porosity and thickness, resulting in an 8.75% increase in power density and significant improvement in oxygen distribution uniformity. To validate the effectiveness and accuracy of the optimization, the optimized structural parameters were incorporated into CFD simulations. The validation results demonstrated close alignment between the optimized model's performance and actual values, confirming the efficacy and reliability of the optimization framework. Overall, this data-driven optimization approach provides an effective tool for multi-variable optimization of complex systems and holds significant importance in enhancing the performance and power density of PEMFC, while also taking into account the biomechanical factors that influence its long-term operation and stability.