An extreme gradient boosting-based thermal management strategy for proton exchange membrane fuel cell stacks

The working temperature is one of the critical factors for proton exchange membrane fuel cell (PEMFC) stacks, as it directly influences the performance and working life of PEMFCs. Due to the inherent nonlinearity of the PEMFC stack thermal management system model, current thermal management strategies (TMSs) generally face with the drawbacks of lower accuracy and robustness. In this research, a novel machine learning (ML) algorithm, i.e. the extreme gradient boosting (Xgboost) algorithm is applied to the TMS of a PEMFC stack to control the inlet and outlet temperatures of the PEMFC stack, where the proton exchange membrane (PEM) water content is also considered and kept at a reasonable level. In order to evaluate the effectiveness of the proposed strategy, a fuel cell hybrid vehicle (FCHV) model from Autonomie software is selected and the proposed TMS is tested and compared to other strategies under a mixed driving cycle for the PEMFC stack. Results show that the proposed Xgboost-based TMS presents the best control performance, which reduces the maximum deviation of the PEMFC stack temperature and the variation range of the PEM water content compared to other ML-based TMSs and the fuzzy logic-based TMS.