A Bayesian framework for revealing degradation mechanisms and early-warning of safety risks in salt spray–aged Lithium-Ion batteries

As the shipping industry transitions towards low-carbon operations, lithium-ion batteries serve as the core energy storage units in all-electric vessels, where their safety directly determines the reliability of both electrical and propulsion systems. However, the high chloride concentration and humidity in the marine environments pose critical challenges for battery safety. In this study, a Bayesian parameter identification framework based on an enhanced single particle model electrochemical model is developed to infer electrochemical parameters with quantified posterior uncertainty for batteries aged under salt spray conditions, achieving a voltage mean absolute error below 12 mV. The inferred parameter trajectories enable the identification of degradation pathways under salt spray ageing, revealing that salt spray exposure significantly alters interfacial reaction kinetics, particularly at the graphite anode. Furthermore, the online safety risk early-warning approach is proposed for salt spray-aged batteries by monitoring abnormal fluctuations in parameters and capacity decay, enabling dynamic risk level assessment. The effectiveness of the proposed approach is validated through the combination of macroscopic observations and microscopic morphological characterization.