This paper proposes a data-driven model for capacity estimation of lithium-ion (Li-ion) battery cells via a temperature-coupled neural network. The model utilizes the temperature measurement of the cell to predict its capacity. In particular, the differential temperature and temperature rise measurements from commercial Li-ion battery cells are employed as model inputs to estimate the cell’s capacity over multiple cycles, until a 30% capacity degradation is observed. The paper presents the model’s development alongside experimental validation to verify its performance.
