State-of-health estimation for lithium-ion batteries based on historical dependency of charging data and ensemble SVR

State of health estimation of LIB based on discharge section with multi-model combined

Accurate estimation of a battery's state of health (SOH) is essential in battery management systems (BMS). This study considers a complete analysis of combining incremental capacity (IC), differential thermal voltammetry (DTV), and differential temperature (DT) for SOH prediction in cases of discharge. Initially, the IC, DTV, and DT curves were derived from the current, voltage, and temperature datasets, and these curves underwent smoothing through the application of Lowess and Gaussian techniques. Subsequently, discerning healthy features were identified within the domains where the curve exhibited substantial phase transitions. Utilizing Pearson correlation analysis, features exhibiting the utmost correlation with battery capacity degradation were singled out. Finally, the state-of-health (SOH) prediction model was constructed using a bidirectional long short-term memory (BILSTM) neural network. Two datasets were used to validate the model, and the experimental results demonstrated that the SOH prediction had a root mean square error (RMSE) below 1.2% and mean absolute error (MAE) below 1%, which verified the feasibility and accuracy. This approach quantifies the internal electrochemical reactions of a battery using externally measured data, further enabling early SOH predictions.

Deep learning to estimate lithium-ion battery state of health without additional degradation experiments

State of health is a critical state which evaluates the degradation level of batteries. However, it cannot be measured directly but requires estimation. While accurate state of health estimation has progressed markedly, the time- and resource-consuming degradation experiments to generate target battery labels hinder the development of state of health estimation methods. In this article, we design a deep-learning framework to enable the estimation of battery state of health in the absence of target battery labels. This framework integrates a swarm of deep neural networks equipped with domain adaptation to produce accurate estimation. We employ 65 commercial batteries from 5 different manufacturers to generate 71,588 samples for cross-validation. The validation results indicate that the proposed framework can ensure absolute errors of less than 3% for 89.4% of samples (less than 5% for 98.9% of samples), with a maximum absolute error of less than 8.87% in the absence of target labels. This work emphasizes the power of deep learning in precluding degradation experiments and highlights the promise of rapid development of battery management algorithms for new-generation batteries using only previous experimental data.