Direct methanol fuel cells parameter identification with enhanced algorithmic technique

• Improving the Artificial Humming Bird Algorithm for better parameter optimization in DMFC fuel cell stacks in real-world use. • Validating the effectiveness of the EAHA on real-world DMFC units by applying it to a specific model. • Demonstrating EAHA's superiority by comparing it to methods like PSO, DA, HHO, RSA, and AHA algorithms. The parameter of a direct methanol fuel cell (DMFC) can be identified using optimization techniques to determine the optimal unknown parameter values that are needed for creating an accurate fuel cell performance prediction model. This research is motivated by the fact that parameter identification process is necessary since manufacturers’ datasheets may not always provide these parameters to users. To address this, the study investigates five optimization techniques with the proposed algorithm for estimating these parameters in DMFCs. A number of optimization techniques have been adopted: Particle Swarm Optimization (PSO), Dragonfly Algorithm (DA), Harris Hawk Optimizer (HHO), Rhinoceros Search Algorithm (RSA), Artificial Hummingbird Algorithm (AHA), and its enhanced version, Enhanced artificial hummingbird algorithm (EAHA). The objective of each approach is to minimize the error between the predicted and measured voltages of the cell and the six unknown parameters. The numerical results support the improvement in the performance and robustness of the proposed approach over the existing methods and the state-of-the-art optimizers. The study shows that the EAHA significantly outperforms other optimization algorithms with a maximum estimation error of 6.34 × 10 −10 . This demonstrates superior performance in finding the global minimum, consistent results with limited variability, and the ability to consistently provide near-optimal solutions. EAHA also exhibits the least standard deviation (2.10 × 10 −10 ), highlighting its reliability and predictability due to EAHA’s efficient balance of exploration and exploitation phases, enabling it to effectively locate and converge on optimal solutions.