A new time-adjustable model-based method for fast open-circuit voltage estimation of Lithium-ion cells

Any system ensuring the safety and the intelligent management of Li-ion batteries requires a thorough knowledge of the curve relating their open circuit voltage to their state-of-charge. However, classical industrial methods used to characterize this relationship necessitate a complete charge/discharge cycle at low current, or with very long relaxation times, to limit the influence of the lithium diffusion. Indeed, the latter induces voltage drops of the same order of magnitude as those induced by the conduction and charge transfer phenomena, but with much higher time constants, generally of several hours. These methods are therefore time-consuming and costly. In this paper, we propose to take advantage of both a continuous-time dynamic model of the Li-ion cell and a dedicated moving horizon estimation algorithm to significantly reduce the characterization time, while simultaneously providing the incremental capacity analysis health indicator. The proposed model structure offers the best trade-off between accuracy of the system simulation and number of parameters that can be sufficiently sensitized regarding the limited duration of the characterization protocol. The originality of the moving horizon estimation algorithm consists in centering the moving observation windows on the current time, improving de facto the estimation accuracy. When using real experimental signals of LFP and NCA cells, we obtain an estimation error lower than 5 mV, close to the best-in-class relaxation method classically used by industrial actors, while dividing the whole experiment time by more than 30.