Improved singular filtering-Gaussian process regression-long short-term memory model for whole-life-cycle remaining capacity estimation of lithium-ion batteries adaptive to fast aging and multi-current variations

For the development of low-temperature power systems in aviation, the transport synergistic carrier optimization of lithium-ions and electrons is conducted to improve the low-temperature adaptability of lithium-ion batteries. In this paper, an improved robust multi-time scale singular filtering-Gaussian process regression-long short-term memory (SF-GPR-LSTM) modeling method is proposed for the remaining capacity estimation. The optimized multi-task training strategy is constructed for the rapid battery performance evaluation, realizing the refined mathematical dynamic characterization for the mapping relationship of the physical carrier transports to obtain the simultaneous improvement of multi-dimensional physical features and a spiral-up iterative optimization scheme. The adaptability of the model is verified using datasets from the whole-life-cycle test conducted on two batteries, and evaluated by root mean squared error (RMSE), mean absolute error (MAE), mean absolute percentage error (MAPE), and R-squared metrics. Even when only using 55% of the datasets to estimate the remaining capacity, the estimation has good effects, with RMSE of 2.3484%, MAE of 0.82526%, MAPE of 0.90716%, and R-squared value of 92.457%. The proposed SF-GPR-LSTM model enables the carrier transport synergistic optimization effectively, laying a theoretical foundation for the whole-life-cycle battery remaining capacity estimation at extremely low temperatures.