A Prediction Model of State of Health for Vanadium Redox Flow Batteries Based on Variational Mode Decomposition and Integrated Neural Network

Vanadium redox flow batteries (VRBs) face the challenge of abnormal capacity degradation due to electrolyte volume imbalance when used for long term energy storage, so it is critical to accurately predict the state of health (SOH) of the batteries to maintain stable operation of the system. In this article, we model the capacity degradation process of VRB and propose the application of variational mode decomposition to SOH time series as a means of addressing the capacity regeneration problem during battery aging. The fluctuation function F(t), which represents capacity regeneration, and the main trend function M(t), which represents the main capacity trends are reconstructed based on correlation analysis. The long short-term memory and the gate recurrent unit are employed to build an integrated neural network model for the properties of the two functions, respectively. The issue of uncertainty of results is solved by calculating probability distributions. The feasibility and validity of the proposed integrated model are verified by experimental and simulation data, respectively. The results demonstrate that the predicted root mean square error of the integrated model can be maintained within 0.45% across multiple time scales. Compared with other models, the proposed model has significant advantages in terms of accuracy and stability.