Answers are grounded in this paper and cite their sources.
Estimating the state of health (SoH) of electric vehicles (EVs) is essential for tackling issues including battery aging, changing circumstances, and a lack of real-time data. Prediction accuracy is improved by data-driven methods, which also ensure performance, longevity, and safety. Therefore, this paper presents a survey of hybrid data-driven techniques for lithium-ion battery (LIB) SoH estimation, focusing on the recurrent neural network and long short-term memory (RNN-LSTM) algorithm, the convolution neural network and long short-term memory (CNN-LSTM) algorithm, the recurrent neural network and bidirectional long short-term memory (RNN-biLSTM) algorithm, and the bidirectional convolutional neural network and long short-term memory (biRNN-LSTM) algorithm. The MATLAB/Simulink platform was utilized to validate this study, and the simulation results demonstrated the accuracy of SoH estimation in terms of mean accuracy error (MAE) and root mean square Error (RMSE). Regarding SoH estimation, MAE and RMSE observations for RNN-LSTM are (0.025, 0.032), for CNN-LSTM are (0.015, 0.026), for RNN-biLSTM are (0.010, 0.015), and for biRNN-LSTM are (0.012, 0.024), respectively. Because of its intricate structure, the RNN-BiLSTM model requires a maximum of 2000 training generations to achieve convergence, in contrast to 1500 for biRNN-LSTM, 1000 for CNN-LSTM, and 500 for RNN-LSTM. In this connection, it has been concluded that the RNN-biLSTM technique is more accurate than the biRNN-LSTM algorithm, which is followed by the CNN-LSTM and RNN-LSTM algorithms.
No takes yet. Share an insight, caveat, or question.
Repalle et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: