Li-ion battery voltage curve reconstruction using partial charge profiles: Actual v/s truncated data

Abstract

Voltage reconstruction is a common technique used in estimation of degradation modes for aged Li-ion batteries. For real-life implementation, it is desirable for voltage reconstruction to work for partial charging as real-life batteries are rarely charged fully. In this pursuit, the presented work investigates a common practice of using truncated data from full charge as a representation of partial charging in voltage reconstruction. Usage of truncated data is prevalent despite known deviations between data collected from partial and full charge cycles and has resulted in a misconception that accurate voltage reconstruction is achievable using partial charging data. Therefore, voltage reconstruction errors between models parametrised using truncated data and actual partial charging data were compared. Results show a four-fold increase in error when using truncated data, which indicates that truncated data is an inappropriate proxy of partial charge. The findings also imply that partial charging is a limitation of voltage reconstruction modelling not highlighted before due to usage of truncated data. This limitation must be addressed to improve the applicability of voltage reconstruction. The study also emphasises the need to generate new battery degradation datasets with appropriate inclusion of partial charging data to enable the development of accurate and holistic models.