Hybrid HGRN-SCSO technique for enhanced prediction of remaining useful life in EV batteries

Predicting battery health is important for ensuring the safety and efficiency of electric vehicle batteries. Existing methods often fail to forecast the remaining useful life, state of charge, and state of health of lithium-ion batteries with high accuracy. This paper developed a new hybrid approach, combining hierarchically gated recurrent neural network and sand cat swarm optimization techniques. The main goal is to improve the prediction of remaining useful life, state of charge, and state of health for lithium-ion batteries in electric vehicles. The hierarchically gated recurrent neural network is used for remaining useful life prediction, while sand cat swarm optimization techniques optimizes its parameters. Using the MATLAB/Simulink working environment, the proposed methods performance is evaluated and contrasted with many existing methods. Compared to existing techniques like artificial neural network and adaptive neuro-fuzzy inference systems, the proposed method achieves lower errors in both charging and discharging conditions. The proposed achieves a mean absolute error of 0.46%, root-mean-square error of 0.56%, and mean absolute percentage error of 0.4%, outperforming adaptive neuro-fuzzy inference system and artificial neural network. Similarly, during discharging, it shows improved performance with root-mean-square error of 0.51%, mean absolute error of 0.42%, and mean absolute percentage error of 0.4%. The proposed method also demonstrates superior accuracy, achieving 97.2% and exhibiting higher sensitivity at 2 kHz. This indicates that the proposed method offers enhanced accuracy and efficiency for battery management in electric vehicles.