State of Charge Estimation of All-Vanadium Redox Battery Based on BIGRU-UKF Network

All-vanadium redox battery (VRB) is an innovative energy storage technology extensively utilized in renewable energy integration and power stabilization, employing vanadium ions as the electrolyte. It offers advantages such as inherent safety, longevity, design flexibility, recoverable capacity, and environmental sustainability. This research proposes a novel VRB model utilizing BIGRU-UKF neural network, which integrates bidirectional time series learning and adaptive noise filtering to enhance the accuracy and real-time performance of state of charge (SOC) prediction greatly. Under changeable flow conditions, the model has a low mean absolute error (MAE) of 0.1312%, a low mean square error (MSE) of 0.0181%, and a coefficient of determination (R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>) of 0.9963, indicating extremely good accuracy. The study presented in this publication offers a novel approach to simulating VRB.