Prediction of future capacity and internal resistance of Li-ion cells from one cycle of input data

Abstract

There is a large demand for models able to predict the future capacity retention and internal resistance (IR) of Lithium-ion battery cells with as little testing as possible. We provide a data-centric model accurately predicting a cell’s entire capacity and IR trajectory from one single cycle of input data. This represents a significant reduction in the amount of input data needed over previous works. Our approach characterises the capacity and IR curve through a small number of key points, which, once predicted and interpolated, describe the full curve. With this approach the remaining useful life is predicted with an 8.6% mean absolute percentage error when the input-cycle is within the first 100 cycles.

Keywords

• Capacity degradation
• Internal resistance degradation
• Prediction of full degradation curve
• Knee and elbow-points
• Lithium-ion cells
• Machine learning