Open-circuit voltage curve reconstruction for degrading lithium-ion batteries utilizing discrete curve fragments from an online dataset

A complete open-circuit voltage (OCV) curve plotted against the state of charge (SOC) for degrading batteries, as a core indicator for battery state estimation and health diagnostics, is very important for whole-life management of battery. Unfortunately, such a curve is almost impossible to obtain in online battery management systems. Due to the uncontrollable OCV sample opportunities, only a series of isolated curve fragments consisting of scarce and discrete OCV-SOC points can be collected. Due to the unavoidable SOC estimation error, the relative position between fragments is uncertain. In order to reconstruct a complete OCV-SOC curve utilizing these isolated OCV curve fragments, an online and training-free curve reconstruction method is developed in this paper. Using this method, all isolated fragments from an online dataset are adaptively rearranged and uniquely located based on the positional interlock between different fragments, and fragments with abnormal state of health (SOH) or measurement errors are screened out. The test results demonstrated that the reconstruction method has a good stability, rapidity, and accuracy. The root mean square error of the curve reconstruction is well controlled within 5 mV throughout the battery's entire lifetime. [Display omitted] •A degrading OCV-SOC curve is reconstructed using only discrete fragments.•Only online operating data without any training experiment are needed.•Good stability, rapidity, and accuracy are verified by experiments.•A high performance and low complexity are simultaneously ensured.•Abnormal and error data will be adaptively labeled and screened out.