In this paper, a novel membrane reactor (MR) for methanol steam reforming is modeled to produce fuel cell grade hydrogen, which can be used as the inlet fuel for a later developed 500-W horizon polymer electrolyte fuel cell (PEFC) stack. The backpropagation (BP) neural network algorithm is employed to develop the mapping relation model between the MR's prime operational parameters and fuel cell output performance for future integration system design and control application. Simulation results showed that the MR model performs well for hydrogen production and the developed PEFC system presents good agreement with experimental results. Finally, the BP method captures an accurate mapping relation model between the MR inputs and PEFC output, for example, predicts the system's behavior well.

中文翻译：

---

基于人工神经网络算法的聚合物电解质燃料电池堆甲醇蒸汽重整器系统行为预测

在本文中，一种用于甲醇蒸汽重整的新型膜反应器 (MR) 被建模以生产燃料电池级氢气，该氢气可用作后来开发的 500-W 水平聚合物电解质燃料电池 (PEFC) 堆的入口燃料。采用反向传播（BP）神经网络算法开发MR主要运行参数与燃料电池输出性能之间的映射关系模型，用于未来集成系统设计和控制应用。仿真结果表明，MR 模型在制氢方面表现良好，所开发的 PEFC 系统与实验结果吻合良好。最后，BP 方法捕获了 MR 输入和 PEFC 输出之间的准确映射关系模型，例如，很好地预测了系统的行为。