State of Health Estimation of LiCoO2 Cells based on Impulse Response and ARMAX Identification

Lithium-ion batteries (LiBs) are subjected to different degradation mechanisms due to storage and operating conditions. State-of-health estimation (SOH) is crucial for proper prediction of battery aging and is usually related to the decrease in energy (capacity fade) or the increase in internal resistance (power fade). This paper evaluates the application of the Impulse Response (IR) method to LiCoO2 (LCO) battery cells for the estimation of SOH in terms of capacity fade. The IR method involves creating a precalculated table in which the cell voltage responses to input pulse currents are measured and stored for different states of charge (SOC) and SOH levels. To this aim, an LCO cell was cycled at constant current steps under fixed-temperature conditions. After a certain number of aging cycles, its voltage response to a current impulse was recorded at different levels of SOC. The same procedure was performed for different SOH levels to train multiple linear auto-regressive identification models. To validate the methodology and assess its precision, the trained models were used to predict the actual SOH level of the cell.