Reinforcement Learning-Based Cell Balancing for Electric Vehicles

Sammanfattning

Lithium-ion battery packs are comprised of hundreds to thousands of individual cellswhich, even though manufactured uniformly, exhibit small variations in their characteristicsthat impact their behavior during operation. These differences cause cells’ Stateof Charge (SOC) to become unbalanced, which can, in turn, reduce the capacity utilizationefficiency of the pack [1]. Additionally, battery cells age differently over time,and fast-aged cells can cause packs with healthy cells to be retired early, without fullytaking advantage of each cell. When a battery has deteriorated to around 80% of itstotal capacity, it is retired from electric vehicle usage [2].To maintain batteries functioning correctly, cell SOC balancing must be done on batterypacks. However, balancing the SOC of cells provides a window of opportunity to alsoinclude cells’ health into the balancing equation, aiming for the homogenization of cellaging, allowing to thoroughly utilize a battery’s resources. In this way, it is possible toboth keep batteries in operating condition and potentially increase their lifespan.In this work, we develop and research a multi-cell simulation framework and ReinforcementLearning (RL) methodologies to explore the potential of cell SOC and healthbalancing. We propose an active balancing strategy for re-configurable cell topologywith RL, in which instead of transferring energy between high SOC cells to low SOCcells, cell utilization is modulated so that the power consumption is optimally distributedbased on each cell’s SOC. This strategy is applied to SOC balancing, as well as SOC andState of Health (SOH) balancing simultaneously, to potentially allow for an exhaustiveutilization of the battery’s potential.