Advancing Electrochemical Modelling of PEM Electrolyzers through Robust Parameter Estimation with the Weighted Mean of Vectors Algorithm

The electrochemical modelling of proton exchange membrane electrolyzers (PEMEZs) relies on the precise determination of several unknown parameters. Achieving this accuracy requires addressing a challenging optimization problem characterized by nonlinearity, multimodality, and multiple interdependent variables. Thus, a novel approach for determining the unknown parameters of a detailed PEMEZ electrochemical model is proposed using the weighted mean of vectors algorithm (WMVA). An objective function based on mean square deviation (MSD) is proposed to quantify the difference between experimental and estimated voltages. Practical validation was carried out on three commercial PEMEZ stacks from different manufacturers (Giner Electrochemical Systems and HGenerators™). The first two stacks were tested under two distinct pressure-temperature settings, yielding five V–J data sets in total for assessing the WMVA-based model. The results demonstrate that WMVA outperforms all optimizers, achieving MSDs of 1.73366e−06, 1.91934e−06, 1.09306e−05, 6.18248e−05, and 4.41586e−06, corresponding to improvements of approximately 88%, 82.9%, 82.4%, 54.5%, and 59.5% over the poorest-performing algorithm in each case, respectively. Moreover, comparative analyses, statistical studies, and convergence curves confirm the robustness and reliability of the proposed optimizer. Additionally, the effects of temperature and hydrogen pressure variations on the electrical and physical steady-state performance of the PEMEZ are carefully investigated. The findings are further reinforced by a dynamic simulation that illustrates the impact of temperature and supplied current on hydrogen production. Accordingly, the article facilitates better PEMEZ modelling and optimizing hydrogen production performance across various operating conditions.