Fuel cells (FCs) have garnered significant attention due to their versatile applications, but their nonlinear characteristics pose challenges in the modeling process. This research presents a unique enhanced artificial hummingbird algorithm (EAHA) aimed at identifying the seven unknown parameters of the proton exchange membrane fuel cells (PEMFCs) stack by utilizing their experimental datasets. To accomplish this, the objective is to achieve accurate current/voltage (I/V) curves where a cost function is defined using the aggregation of quadratic deviations (AQD) between the measured dataset points and the appropriate model-based estimations. The presented EAHA combines several territorial foraging techniques with a linear regulating mechanism. The performance of the conventional AHA is compared with the suggested EAHA using three commonly employed PEMFC modules. Furthermore, a comparative analysis is conducted against previously published methodologies and newly developed optimizers such as the equilibrium optimizer (EO), social networking search (SNS) technique, slim mold algorithm (SMA), heap-based optimizer (HBO), and African vultures optimization (AVO) technique. The findings are compared to existing methodologies and other state-of-the-art optimizers, providing valuable insights into the efficacy of the proposed approach. For the 250 W PEMFC stack, the proposed EAHA shows improvements of 2.966%, 6.493%, 1.491%, 7.080%, 1.131%, and 2.875% over AHA, AVO, EO, HBO, SMA, and SNS, respectively, depending on the mean AQD values. Similar findings are attained for the other two stacks. For example, for the test case of the BCS 500 W PEMFCs stack, the proposed EAHA demonstrates improvements of 64.228%, 82.859%, 66.140%, 81.156%, 46.302%, and 71.635% over AHA, AVO, EO, HBO, SMA, and SNS, respectively.