Prediction and Optimization of Bolt Parameters for Automotive Fuel Cell Stack under Random Vibration

Proton exchange membrane fuel cells (PEMFCs) directly convert chemical energy into electrical energy, offering high energy conversion efficiency, zero emissions, low noise, and rapid response, making them highly promising for vehicular applications. However, during vehicle operation, PEMFC stacks are inevitably subjected to random vibration loads due to road unevenness and internal system vibrations. These mechanical disturbances can induce structural displacement and deformation within the fuel cell stack, significantly affecting its durability and operational safety. Thus, this study conducts a random vibration analysis of fuel cell stacks under varying fastening bolt parameters. A dataset is constructed, comprising bolt parameters and the corresponding stress responses, which is then used to train an extreme gradient boosting (XGBoost) surrogate model. By integrating XGBoost with a genetic algorithm (GA), a GA-XGBoost predictive model is developed to optimize the structural parameters of the fastening bolts. The optimization results indicate that the optimal bolt diameter is 5.5 mm, the optimal nut thickness is 6.5 mm, and the optimal bolt-to-stack side distance is 10 mm. The predicted minimum stress is 52.351 Pa, representing a reduction of 2.849 Pa compared to the lowest stress observed in the dataset, thereby enhancing the structural durability of the fuel cell stack.