In battery energy storage systems, accurately estimating battery capacity and state of health is crucial to ensure satisfactory operation and system efficiency and reliability. However, these tasks present particular challenges under irregular charge–discharge conditions, such as those encountered in renewable energy integration and electric vehicles, where heterogeneous cycling accelerates degradation. This study introduces a hybrid deep learning framework to address these challenges. It combines convolutional layers for localized feature extraction, bidirectional recurrent units for sequential learning and a temporal attention mechanism. The proposed hybrid deep learning model, termed CBATE-Net, uses ensemble averaging to improve stability and emphasizes degradation-critical intervals. The framework was evaluated using voltage, current and temperature signals from four benchmark lithium-ion cells across complete life cycles, as part of the NASA dataset. The results demonstrate that the proposed method can accurately track both smooth and abrupt capacity fade while maintaining stability near the end of the life cycle, an area in which conventional models often struggle. Integrating feature learning, temporal modelling and robustness enhancements in a unified design provides the framework with the ability to make accurate and interpretable predictions, making it suitable for deployment in real-world battery energy storage applications.

Data-driven battery state estimation has been widely studied in recent years to enable safe, reliable, and lifetime-aware battery management in e-mobility applications, where accurate estimation is essential under highly dynamic operating conditions. With several recent reviews focusing on model architectures and methodological developments, a critical assessment of practical feasibility, real-world deployability, and application-specific challenges in e-mobility is needed. This review therefore provides a challenge-oriented assessment of recent data-driven methods for battery state estimation, with particular emphasis on practical deployment in e-mobility systems. Beyond summarising existing approaches, it assesses how recent studies address key deployment barriers. These barriers include data dependency, limited ground-truth health labels, and fragmented datasets. Further challenges concern feature complexity and interpretability, robustness under domain shifts, embedded-BMS constraints, and cell-to-pack scalability.Recentresearch increasingly focuses on battery health analytics, reflecting the importance of ageing-aware battery management in e-mobility. Based on the gaps identified, future research directions are outlined to support the development of safe, reliable, interpretable, and deployment-ready battery state estimation for next-generation e-mobility systems.

Multilevel active bridge converters are potential candidates for many modern high-power DC applications due to their ability to integrate multiple sources while minimizing weight and volume. Therefore, this paper deals with an analytical, simulation-based, and experimentally verified investigation of their circulating current behavior, power factor performance, and power loss characteristics. A high-frequency link analysis framework is developed to characterize voltage, current, and power transfer waveforms, providing insight into reactive power generation and its impact on overall efficiency. By introducing a modulation-based control approach, the proposed converters significantly reduce circulating currents and enhance the power factor, particularly under varying phase-shift conditions. Compared to quadruple active bridge topologies, the discussed multilevel architectures offer reduced transformer complexity and improved power quality, making them suitable for demanding applications such as electric vehicles and aerospace systems.