Precise models predicting fuel cell performance under different operating conditions require accurate parameter identification in a proton exchange membrane fuel cell (PEMFC). Most traditional parameter estimation methodologies depend on optimization algorithms which are limited in their efficiency, convergence speed, and robustness. Typically, existing algorithms fail to achieve a balance between precision and computational efficiency, leading to suboptimal modeling of the complex, nonlinear behavior of PEMFCs. In this paper, we present the two-stage differential evolution (TDE) algorithm, which fills these gaps by using a new mutation strategy that improves solution diversity and speeds up convergence. Seven critical unknown parameters ([Formula: see text] and λ) in PEMFC models are identified by using the proposed TDE algorithm. The optimization process is to minimize the sum of squared errors (SSE) between the experimentally measured and predicted cell voltages. TDE resulted in a 41% reduction in SSE (minimum SSE of 0.0255 compared to 0.0432), a 92% improvement in maximum SSE, and over 99.97% reduction in standard deviation compared to the HARD-DE algorithm. Furthermore, TDE was shown to be 98% more efficient than HARD-DE, with a runtime of 0.23 s, compared to HARD-DE's runtime of 11.95 s. Extensive testing of these advancements was performed on six commercially available PEMFC stacks over twelve case studies, and I/V and P/V characteristics were confirmed to be consistent with experimental data. The results show that TDE has better accuracy, robustness and computational efficiency than the other methods, and therefore TDE can be used as a real time PEMFC parameter estimation tool.