Abstract The fast and precise automated assembly of limp components to large stacks is a technical challenge. For high voltage fuel cell stacks, hundreds of thin, limp and brittle parts must be stacked precisely. To ensure a robust and fast stacking process, a deviation compensation strategy is presented which allows for increased precision and accuracy through modeling of process-specific deviations. Potential multidimensional regression methods for modeling such deviations are compared. Supported by artificial neural networks, extensive handling operations are performed by a robot-based fuel cell stacking system. The results are statistically evaluated and discussed with regard to precision.

中文翻译：

---

由人工神经网络支持的柔软燃料电池组件的快速精确拾取和放置堆叠

摘要 将柔软的组件快速、精确地自动组装到大型堆栈是一项技术挑战。对于高压燃料电池堆，必须精确堆叠数百个薄的、柔软的和易碎的部件。为确保稳健且快速的堆叠过程，提出了一种偏差补偿策略，该策略通过对过程特定偏差的建模来提高精度和准确度。比较了用于模拟这种偏差的潜在多维回归方法。在人工神经网络的支持下，大量的搬运操作由基于机器人的燃料电池堆叠系统执行。对结果进行统计评估并就精度进行讨论。