Machine learning-enabled performance prediction and optimization for iron–chromium redox flow batteries
Yingchun Niu, Ali Heydari, Wei Qiu, Chao Guo, Yinping Liu, Chunming Xu, Tianhang Zhou, Quan Xu
Abstract
> 0.92) to link ICRFB properties to energy efficiency, coulombic efficiency, and capacity. We also interpret the ML models based on Shapley additive explanations and extract valuable insights into the importance of descriptors. It is noted that the operation conditions (current density and cycle number) and the electrode type are the most critical descriptors affecting the voltage efficiency and coulombic efficiency while the electrode size strongly affects the capacity. Moreover, active learning is used to explore the most optimized cases considering the highest energy efficiency and capacity. The versatility and robustness of the approach are demonstrated by the successful validation between ML prediction and our experiments of energy efficiency (±0.15%) and capacity (±0.8%). This work not only affords fruitful data-driven insight into the property-performance relationship, but also unveils the explainability of critical properties on the performance of ICRFBs, which accelerates the rational design of next-generation ICRFBs.