Online determination of vanadium concentration and SOC of positive electrolyte in VFBs by machine learning based spectroscopic study

With the increasing severity of energy crises and environmental pollution, the development of efficient energy storage technologies to address the intermittent and unstable nature of renewable energy has become a critical research focus. Vanadium flow batteries (VFBs), owing to their high safety, long cycle life, and large-scale energy storage capacity, have emerged as a promising candidate for grid-scale energy storage. However, practical operation of VFBs faces challenges such as electrolyte concentration imbalance, self-discharge, and side reactions, necessitating online detection techniques for real-time monitoring of electrolyte states. This study proposes an online detection method integrating machine learning with spectral analysis to simultaneously determine vanadium ion concentration and state of charge (SOC) in the positive electrolyte of VFBs. By establishing a standardized spectral database and training machine learning models, high-precision predictions of vanadium concentration and SOC were achieved. Experimental results demonstrate that the optimal model achieves relative prediction errors within 0.28 % for vanadium concentration and 2.30 % for SOC, indicating robust practicality and reliability. Furthermore, long-cycle tests have verified the effectiveness of this method in real-time monitoring of electrolyte concentration and SOC variations. This study provides important technical support for optimizing the performance and operational management of vanadium flow batteries while also offering a potential approach for further investigating the mechanisms of vanadium ion transmembrane migration.