Battery prognostics based on discharge voltage drop for energy storage applications

Abstract

This study proposes an approach that can predict the end of Li-ion battery life using the discharge voltage drop curve during its use in the energy storage system (ESS). The approach is developed based on the findings that the voltage drop in Li-ion batteries increases as the battery undergoes cycles, and it can be related with the residual capacity. The key idea is to insert the additional cycle of full charging and discharge with constant c-rate during the usage of the ESS. In this cycle, the relation between the voltage drop and capacity is established off-line via regression technique. Then this is applied to estimate the SOH and RUL on-line during the battery cycles. Particle filter (PF) algorithm is applied to this end, in which the degradation and regression models are taken as the state and measurement models respectively, and the capacity is estimated in the form of samples. The obtained samples are then used to predict its behavior in the future, from which the RUL distribution is determined. Conclusion of the study is that the voltage drop in Li-ion batteries can be a good indicator of the battery health and PF is a useful toolthat can predict the RUL accurately even when the chargedischarge conditions change in the middle of the usage cycles.