Comparative Analysis of Rule-Based and Model Predictive Control Algorithms in Reconfigurable Battery Systems for EV Fast-Charging Stations

Electric vehicles (EVs) can be fast-charged without inducing abrupt loads on the grid if stationary battery systems are used as buffers. The investigated commercial 300-cell reconfigurable battery system (RBS) does not need a DC/DC converter to connect to the EV or the electrical grid, but can set the required voltage/current with engaging or bypassing individual cells. Stacking the cell voltages introduces the problem of voltage granularity, which causes current ripples during the reconfiguration. Furthermore, since the battery cells in question live their second life application, their parameters are heterogeneous. It is necessary to balance the state-of-charge (SoC) and the state-of-health (SoH) of the battery cells to provide a robust capacity, power and lifetime for the system. The present work compares a rule-based control (RBC) and a model predictive control (MPC) algorithm for battery management. The control objectives consist of tracking the constant current-constant voltage (CCCV) reference, minimizing the current ripple, and balancing the SoC and the SoH of the cells. The RBC is inspired by an industrial battery management system, while the MPC solves a mixed-integer-linear-program (MILP) problem. A high fidelity stochastic simulation environment is presented, and the online performance and the applicability of the control algorithms are compared via Monte Carlo method. In the simulation, realistic heterogeneity of cells, cars and probability of car turnups are implemented. The MPC outperforms the RBC on current ripple suppression by 25%, while the performance regarding the rest of the objectives is comparable. The RBC is at least 100 times faster and it is preferable if less time and resource is to be invested, while the MPC is preferable for performance sensitive applications or more complex control objectives.