Fast Model Predictive Control for Redistributive Lithium-Ion Battery Balancing

Energy storage systems with Lithium-ion batteries require balancing due to individual cells having manufacturing inconsistencies, different self-discharge rates, internal resistances, and temperature variations. Nondissipative redistributive balancing further improves on the pack capacity and efficiency over a dissipative approach where energy is consumed across shunt resistors. This paper presents a high-level fast model predictive control (MPC) in continuous time. The optimization problem uses performance metrics to balance the state of charge (SoC) in the battery pack. It is shown in simulation that MPC achieves a single point convergence of the SoC when compared against a common rule-based algorithm. This improves the efficiency of the power electronics and prolongs the life of each battery cell since frequent switching between charging and discharging of intermediate cells is avoided. Experimental results are presented to show a redistributive battery balancing system that achieves a balanced state in the minimum amount of time by coupling the fast MPC with microcontrollers available on todays market.