Hybrid temporal deep learning and ensemble regression framework for remaining useful life prediction of lithium-ion batteries in energy storage systems
Jul 2026· Energy Exploration & Exploitation· 0 citations· 37 references
TL;DR
A hybrid data-driven framework integrating a Temporal Convolutional Network, Bidirectional Long Short-Term Memory, and Extreme Gradient Boosting for accurate LIB RUL prediction provides a robust and computationally efficient solution for intelligent battery health monitoring, predictive maintenance, and smart battery management applications in electric vehicles and energy storage systems.
Abstract
Accurate prediction of the remaining useful life (RUL) of lithium-ion batteries (LIBs) is essential for enhancing the reliability, operational safety, and energy management efficiency of electric vehicles and modern energy storage systems. However, battery degradation is governed by highly nonlinear electrochemical mechanisms and complex temporal dependencies that are difficult to model using conventional physics-based approaches or standalone machine learning techniques. To address these challenges, this article proposes a hybrid data-driven framework integrating a Temporal Convolutional Network (TCN), Bidirectional Long Short-Term Memory (BiLSTM), and Extreme Gradient Boosting (XGBoost) for accurate LIB RUL prediction. The proposed architecture utilizes the TCN module to capture short-term temporal degradation patterns from sequential battery operational data, while the BiLSTM network learns long-term temporal dependencies and degradation evolution across multiple charge–discharge cycles. The deep temporal representations extracted by the TCN–BiLSTM network are subsequently processed using an XGBoost regression model to effectively model nonlinear relationships between battery operational characteristics and RUL. The framework is validated using the NASA LIB aging dataset containing 14,896 charge–discharge cycle samples with operational features including cycle index, discharge time, charging duration, voltage degradation characteristics, and constant-current charging behavior. Statistical analysis and Min–Max normalization are employed to improve feature consistency, numerical stability, and model convergence. Experimental results demonstrate that the proposed framework effectively captures battery degradation dynamics and achieves highly accurate and stable prediction performance. Five-fold cross-validation results yield a low Mean Absolute Error of 0.00957, Root Mean Square Error of 0.02926, and a high coefficient of determination (
R
2
) of 0.9882, indicating excellent predictive capability and strong generalization performance. Comparative analysis further demonstrates that the proposed hybrid framework outperforms conventional Random Forest, XGBoost, LSTM, and BiLSTM models in terms of prediction accuracy and robustness. In addition, ablation analysis confirms the complementary contribution of temporal convolutional learning, sequential dependency modeling, and ensemble nonlinear regression toward improved RUL estimation. The proposed framework provides a robust and computationally efficient solution for intelligent battery health monitoring, predictive maintenance, and smart battery management applications in electric vehicles and energy storage systems.
Lithium-ion battery remaining useful life (RUL) prediction is strongly affected by nonlinear degradation behavior and complex temporal dependence under practical operating environments. To improve prediction accuracy and robustness, this study develops a hybrid prediction framework integrating convolutional neural networks (CNN), bidirectional long short-term memory networks (BiLSTM), attention mechanisms, and ensemble learning strategies. First, multidimensional battery operation signals, including voltage, current, temperature, charging capacity, and discharge energy, are normalized and processed to extract representative degradation information. An attention mechanism is further introduced to adaptively emphasize important degradation stages and suppress irrelevant temporal information. In addition, a stacked ensemble structure combining support vector regression (SVR) and LightGBM is designed to enhance model generalization and reduce prediction variance. Experiments conducted on the CALCE lithium-ion battery dataset demonstrate that the proposed framework achieves excellent predictive performance, with an R² value of 0.9995, RMSE of 0.0504, and MAE of 0.0476. Compared with conventional CNN, LSTM, BiLSTM, and hybrid deep learning approaches, the proposed method exhibits higher prediction stability and stronger adaptability under complicated operating conditions. The developed model provides an effective technical solution for intelligent battery health management and remaining useful life estimation.
Yi-Bao Zhang· European Conference on Elect...· 0 citations
Accurate prediction of the remaining useful life (RUL) of lithium-ion batteries is essential for the reliability and safety of modern energy systems. However, the capacity regeneration (CR) phenomenon, a temporary recovery of capacity during cycling or rest, introduces non-monotonic fluctuations in degradation trajectories, posing significant challenges to existing prediction models. Many current approaches treat CR as noise or overlook its physical significance, limiting interpretability and accuracy. To address this, we propose a hybrid framework that explicitly models both regenerative and degenerative battery behaviors. The method uses Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) to decompose capacity sequences into high- and low-frequency components. A graph convolutional network-long short-term memory (GCN-LSTM) branch captures high-frequency local fluctuations containing regeneration-related information, while a deep neural network (DNN) branch learns the long-term degradation trend. These are fused to reconstruct the full degradation path and predict RUL. The CEEMDAN-GCN-LSTM-DNN hybrid model achieves a mean absolute percentage error below 0.15%, outperforming several state-of-the-art baselines. It also demonstrates strong robustness under data-limited conditions and effectively captures complex CR patterns often missed by conventional methods. This study offers a new perspective for handling non-monotonic battery degradation and provides a useful tool for battery health management and predictive maintenance.
Qiang-Xiang Zhai, Hong-Min Jiang, Yao Lu et al.· Engineering Research Express· 0 citations
Accurate estimation of the state of charge (SOC) and state of health (SOH) of lithium-ion batteries is essential for improving the performance, safety, and lifespan of electric vehicles (EVs). Traditional estimation methods often face challenges such as high computational complexity, limited adaptability to battery aging, and reduced accuracy under varying operating conditions. To overcome these limitations, this paper presents a hybrid data-driven framework that combines machine learning and deep learning techniques for SOC and SOH prediction. For SOC estimation, linear regression, recurrent neural networks (RNN), gated recurrent unit (GRU), and stacked long short-term memory (LSTM) models are employed. For SOH prediction, ensemble learning methods, including stacking regressor, tree-based pipeline optimization tool (TPOT) regressor, and hybrid GRU-LSTM models, are utilized. The proposed models are evaluated using publicly available lithium-ion battery datasets under different charge-discharge conditions. Results show that deep learning approaches achieve superior performance, with GRU-LSTM and stacked LSTM models providing highly accurate SOC estimation (R² ≈ 0.993, RMSE ≈ 0.015), while the TPOT-based ensemble model delivers near-perfect SOH prediction (R² ≈ 1.0). A web-based implementation further enables real-time battery monitoring, demonstrating the framework’s practicality for advanced battery management systems (BMS) in EV applications.
N. Keerthi, B. Jyothi, M. Sharanya et al.· International Journal of App...· 0 citations
Accurate state of health (SOH) estimation of lithium-ion batteries is essential for ensuring the safe and efficient operation of electric transportation and grid-scale energy storage systems (ESS). However, extracting reliable degradation information from Battery Management System (BMS) data remains challenging due to measurement noise, operating variations, and nonlinear characteristics of charging signals. To address this issue, a multi-scale spatiotemporal deep learning framework based on a multi-scale convolutional neural network and bidirectional long short-term memory (MS-CNN-BiLSTM) is proposed. A degradation-aware sliding-window strategy is first designed to extract aging-related features from charging signals, while parallel multi-scale convolution branches with different receptive fields are employed to capture temporal characteristics at multiple scales. Subsequently, an attention-enhanced BiLSTM module is introduced to aggregate long-term degradation dependencies and adaptively capture informative temporal representations. The proposed framework is evaluated using lithium-ion batteries with different chemistries, including the NASA LCO and MOLICEL NCM datasets. Experimental results demonstrate that the proposed method achieves accurate SOH estimation with MAE values as low as 0.0026 and R2 values above 0.979. Furthermore, the model maintains consistent estimation performance across different battery chemistries with relatively low computational complexity, suggesting its potential suitability for embedded BMS applications.
The Energy Storage System (ESS) is an important component of the Electric Vehicle (EV) system, wherein Lithium-ion (Li-ion) batteries are commonly deployed owing to high energy storage capacity and durability. But charging and discharging over time causes reduction in the efficiency of batteries. Further, degradation becomes very fast when batteries reach End of Life (EOL). This implies the need for efficient battery management systems in EVs. The Battery Management System (BMS) monitors various indicators of the battery including State of Charge (SOC), Remaining Useful Life (RUL), and State of Health (SOH). The monitoring of RUL and SOH is especially useful for forecasting the degradation of batteries and thus minimizing maintenance expenses. In this work, a deep learning framework-based approach for predicting the RUL and SOH of Li-ion batteries has been presented. In the initial stage, the acquired health data from the batteries is preprocessed using Min-Max normalization for consistency in data. After that, the prediction process is done using the developed Dream Optimized Explainable Bayesian Gated Recurrent Unit (DO EB_GRU). In this method, the Explainable Bayesian Gated Recurrent Unit (EB_GRU) is optimized using the Dream Optimization Algorithm (DOA). Experimental examination specified that DO EB_GRU accomplished an MSE of 0.0016, MAE of 0.015, R-Squared of 0.956.
Priya A Geevarghese, L. Suresh, Aneesh P. Thankachan· International Conference on...· 0 citations
Lithium-ion batteries are widely used in various energy sectors, and accurately predicting their early remaining useful life (RUL) is crucial for shortening battery evaluation time and accelerating battery commercialization. However, information on degradation during the early cycling stages of batteries is limited, and it is difficult to fully characterize their lifespan. This study proposes a CNN–Transformer–BiGRU-based method for predicting the early RUL of lithium-ion batteries using Grey Wolf Optimization (GWO). First, using only the first 100 cycles of each battery in the MIT dataset, early degradation features are extracted from the dimensions of capacity and internal resistance, and then standardized. Second, a CNN is employed to extract local degradation features, while the Transformer’s self-attention mechanism is used to capture global correlations, and BiGRU is utilized to further extract bidirectional temporal dependency information. Building on this foundation, GWO is introduced to perform joint optimization of the model’s key hyperparameters to obtain optimal network parameters. Finally, the effectiveness of the proposed method is validated through ablation and comparison experiments. The experimental results show that the proposed model achieved an R2 of 0.9633, with RMSE, MAE, and MAPE values of 80.5608 cycles, 63.2524 cycles, and 7.29%, respectively, demonstrating overall prediction performance superior to that of the comparison models. This method can effectively mine degradation information related to battery life from limited early-cycle data, providing an effective approach for the accurate prediction of the early RUL of lithium-ion batteries.
