Online estimation of SOH for lithium-ion battery based on SSA-Elman neural network

Abstract The estimation of state health (SOH) a lithium-ion battery (LIB) is great significance to system safety and economic development. This paper proposes SOH method based on the SSA-Elman model for first time. To improve correlation rates between features capacity, combining median absolute deviation filtering Savitzky–Golay proposed process data. Based aging characteristics LIB, five with above 0.99 after data processing are then proposed. Addressing defects Elman model, sparrow search algorithm (SSA) used optimize network parameters. In addition, incremental update mechanism added generalization model. Finally, performance verified NASA dataset, outputs Elman, LSTM models compared. results show that can accurately estimate SOH, root mean square error (RMSE) being as low 0.0024 percentage (MAPE) 0.25%. RMSE does not exceed 0.0224 MAPE 2.21% in high temperature verifications.