Data-driven state of health and state of safety estimation for alternative battery chemistries — A comparative review focusing on sodium-ion and LFP lithium-ion batteries

Abstract

This paper presents a comprehensive survey on data-driven online estimation of state of health (SoH) for alternative battery chemistries for maritime applications, with a particular focus on LFP lithium-ion and sodium-ion types of batteries. In addition, the emerging concept of state of safety (SoS), a critical yet underexplored metric for maritime battery systems, is explored. Building on previous work on nickel–manganese–cobalt (NMC) lithium-ion batteries, this study evaluates the applicability of existing SoH estimation methodologies to alternative chemistries. The findings suggest that similar data-driven approaches, including empirical and semi-empirical methods, physics-based models, machine learning models, and hybrid approaches, can be employed across these chemistries. However, the methods require calibration, fine-tuning, and validation for each specific battery type. It is believed that SoS holds significant potential for maritime applications, provided it incorporates a relevant set of safety sub-functions with properly defined thresholds and warning criteria. Its integration into real-time monitoring systems appears feasible, given continuous measurement of relevant inputs. However, further research is recommended on how to best account for interdependencies between the various safety sub-function and correlations in the input data as well as how to account for the effect of degradation on SoS. Additionally, it seems reasonable to investigate whether some kind of memory could be incorporated in order to account for the experience of previous abusive conditions.