Chaib Lakhdar, Khemili Fatima Zahra, Tadj Mohammed, Choucha Abdelghani, Namomsa Borchala, Elsayed Salah K, Ghoneim Sherif S M, Abou Sharaf Ahmed B
Energy and Materials Laboratory, University of Tamanghasset, P.O. Box 10034, 11001, Sersouf, Tamanghasset, Algeria.
Department of Electrical and Computer Engineering, Jimma University Institute of the Technology Jit, P.O.Box 378, Jimma Town, Ethiopia.
Sci Rep. 2025 Mar 14;15(1):8802. doi: 10.1038/s41598-025-93162-7.
Proton exchange membrane fuel cells (PEMFCs) have emerged as a promising renewable energy source, generating significant interest in recent years due to their high efficiency, low operating temperature, and durability. Accurately estimating seven unknown parameters in the PEMFC electrochemical model is crucial for developing a more precise model, thereby improving the efficiency and performance of PEMFC systems. For this reason, a new optimization method inspired by parrots' (pyrrhura molinaes') behavior, named Parrot Optimizer (PO), is introduced here to address the problem of optimal parameter identification ([Formula: see text]) in PEMFC models. The estimate of these unknown characteristics is treated as a challenging, nonlinear optimization issue that has to be addressed with a strong optimization technique. The paper outlines two improvements to the basic PO algorithm: the first involves employing Opposition-based Learning to boost the search efficiency and refine candidate solution generation. The second integrates a Local Escaping Operator with PO to boost the exploration capabilities mitigate the risk of getting trapped in local optima, and enhance overall convergence behavior. The IPO was rigorously validated through the application of benchmark functions to assess its performance. Three distinct PEMFC stacks, the NedStackPS6, BCS Stack, and Ballard Mark V, have been used to empirically demonstrate the efficacy of this improved PO in optimizing the PEMFC model. Several recognized modeling approaches from the literature are used in a comprehensive examination to show the method's efficacy and dependability. For the NedStackPS6, BCS Stack, and Ballard Mark V units, the corresponding SQE values are 2.065816 V, 0.012457 V, and 0.814325 V. The IPO demonstrates a 12.87% improvement in the best measure and an 88.37% reduction in standard deviation compared to PO. The results show that the designed approach, including sensitivity analysis, correctly characterizes the PEMFC model. The improved PO effectively achieves the lowest SQE values and consistent convergence trajectories.
质子交换膜燃料电池（PEMFCs）已成为一种很有前景的可再生能源，近年来因其高效率、低工作温度和耐久性而备受关注。准确估计PEMFC电化学模型中的七个未知参数对于开发更精确的模型至关重要，从而提高PEMFC系统的效率和性能。因此，本文引入了一种受鹦鹉（绿颊锥尾鹦鹉）行为启发的新优化方法，即鹦鹉优化器（PO），以解决PEMFC模型中的最优参数识别问题（[公式：见原文]）。对这些未知特性的估计被视为一个具有挑战性的非线性优化问题，必须用强大的优化技术来解决。本文概述了对基本PO算法的两项改进：第一项改进是采用基于对立的学习来提高搜索效率并优化候选解的生成。第二项改进是将局部逃逸算子与PO集成，以提高探索能力，降低陷入局部最优的风险，并增强整体收敛行为。通过应用基准函数对IPO进行了严格验证，以评估其性能。使用了三个不同的PEMFC电池堆，即NedStackPS6、BCS电池堆和巴拉德Mark V，通过实验证明了这种改进的PO在优化PEMFC模型方面的有效性。综合检验中使用了文献中几种公认的建模方法，以展示该方法的有效性和可靠性。对于NedStackPS6、BCS电池堆和巴拉德Mark V单元，相应的SQE值分别为2.065816 V、0.012457 V和0.814325 V。与PO相比，IPO在最佳测量值上提高了12.87%，标准差降低了88.37%。结果表明，所设计的方法，包括灵敏度分析，正确地表征了PEMFC模型。改进后的PO有效地实现了最低的SQE值和一致的收敛轨迹。
