The inconsistency of battery voltages in all-electric ships is a significant issue for electric vehicle battery systems, leading to numerous safety concerns during vessel operation. Therefore, timely fault diagnosis and accurate fault prediction are crucial for the safe operation of ships. This study examines the fault alarm system of marine battery management systems in conjunction with the unique operating conditions of ships, focusing on the system’s latency. To facilitate prompt fault detection, a fault diagnosis method based on the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm is proposed, utilizing the voltage data of battery clusters. Results indicate that the DBSCAN clustering algorithm demonstrates superior effectiveness and accuracy in identifying irregular battery clusters. Furthermore, the fault prediction method based on the iTransformer model is introduced to forecast variations in battery cluster voltages. Experimental findings suggest that this model can effectively predict consistency faults and over-/under-voltage conditions based on battery cluster voltage values and corresponding fault thresholds.

With the widespread application of fuel cell technology in the fields of transportation and energy, Battery Management Systems (BMSs) have become one of the key technologies for ensuring system stability and extending battery lifespan. As an auxiliary power source in fuel cell systems, the prediction of the Remaining Useful Life (RUL) of lithium-ion batteries is crucial for enhancing the reliability and efficiency of fuel cell ships. However, due to the complex degradation mechanisms of lithium batteries and the actual noisy operating conditions, particularly capacity regeneration noise, accurate RUL prediction remains a challenge. To address this issue, this paper proposes a lithium battery RUL prediction method based on an Adaptive Modal Enhancement Network (RIME-VMD-SEInformer). By incorporating an improved Variational Mode Decomposition (VMD) technique, the RIME algorithm is used to optimize decomposition parameters for the adaptive extraction of key modes from the signal. The Squeeze-and-Excitation Networks (SEAttention) module is employed to enhance the accuracy of feature extraction, and the sparse attention mechanism of Informer is utilized to efficiently model long-term dependencies in time series. This results in a comprehensive prediction framework that spans signal decomposition, feature enhancement, and time-series modeling. The method is validated on several public datasets, and the results demonstrate that each component of the RIME-VMD-SEInformer framework is both necessary and justifiable, leading to improved performance. The model outperforms the state-of-the-art models, with a MAPE of only 0.00837 on the B0005 dataset, representing a 59.96% reduction compared to other algorithms, showcasing outstanding prediction performance.