Combined Cell-Level Estimation of State-of-Charge and Temperature in Battery Packs

Abstract

Accurately estimating the State-of-Charge (SoC) and temperature of lithium-ion cells inside a battery pack is critical for safe and reliable operation. This paper extends battery state estimation from single-cell SoC estimation towards a combined SoC and temperature estimation for a multi-cell pack. Combining the electrical and thermal models on a pack level allows for the inclusion of thermal interaction between the cells in a compact manner. The resulting coupled model is used in an extended Kalman filter with correlated noise and a forgetting factor, that is extended with model-residual-based tuning to accommodate differences in magnitude between the electrical and thermal parts of the coupled model. Combined estimation of temperature and SoC decreases the SoC error by around 50%, while the included thermal model adds an accurate estimate of the individual cell temperatures. This provides insight into the temperature distribution, without requiring a large number of temperature sensors.