Nonlinear observability and identifiability of single cells in battery packs

Abstract

Lithium-ion batteries currently offer the best tradeoff between objectives like performance, energy density and lifetime. To serve the demands of many applications, often a large number of single battery cells are combined into modules or packs of batteries. However, the determination of state of charge and parameters of the single cells is important for the reliable and safe operation of batteries. In the present paper, we investigate the observability and identifiability of cells in battery packs, realized in parallel and serial connections. The analysis is based on linear and nonlinear observability tests exploiting an equivalent circuit model. This leads to conclusive findings concerning the feasibility of estimating the states of single cells from lumped measurements. As shown for cells in parallel connection, one voltage and one current sensor are sufficient to determine state of charge and model parameters of the individual cells. The results are illustrated by simulations.