Data-Driven Control for Constant-Temperature Fast Charging of Lithium-Ion Cells

Achieving optimal and fast charging in lithium-ion batteries without compromising safety and lifetime requires accurate models, usually obtained through long and cumbersome cell characterization campaigns. The high complexity of the model, and especially the variations in battery parameters due to aging and other external factors, makes model-based control strategies often unsatisfactory. To overcome this shortcoming, this paper proposes a data-driven approach for constant-temperature constant-voltage (CT-CV) charging of lithium-ion cells, using Data-enabled Predictive Control (DeePC). The DeePC exploits current, temperature and voltage data to optimize charging trajectories based on battery behavior, avoiding the need of a battery model and a state-of-charge (SOC) estimation algorithm. Experimental validation with Samsung INR18650-20R cells reveal superior performance, achieving faster charging compared to conventional CC-CV techniques while maintaining precise thermal regulation in every condition, without the need of a model.