State of Charge Estimation of Sodium-Ion Batteries Based on N-BEATS Neural Network

The increasing penetration of renewable energy sources in the grid and the deficiencies of lithium batteries in terms of storage capacity have necessitated the use of sodium batteries as large scale energy storage devices in renewable energy systems. Accurate estimation of the State of Charge (SOC) of such batteries is of great significance. In this paper, we fabricated the CR2032 sodium-ion batteries and propose a novel method for the estimation of SOC of sodium-based coin cell using a Neural Basis Expansion Analysis for Time Serious (N-BEATS). The advantage of this method is that it does not require the input features to estimate the SOC, instead, the historical information of time series is used to predict the future information. Therefore, there is no need to collect external observations to achieve battery management system (BMS) cost savings. Experimental results show that the proposed method can substantially reduce the estimation error, with a mean absolute error (MAE) and root mean square error (RMSE) of less than 0.3669 % and 0.4549 %, respectively. This improved reliability of the BMS under the energy storage system (EES) is of great significance.