Power Losses Aware Nonlinear Model Predictive Control Design for Active Cell Balancing

Active cell balancing guarantees good performance and long life of a battery pack. In this letter a high-fidelity model, considering the static and dynamic parameters, is developed for computing the balancing currents and power losses of an active cell balancing network (ACBN). The model comprises any two adjacent Li-ion cells connected in series and a bidirectional buck-boost converter. This model is employed to design a nonlinear model predictive controller (NMPC), which minimizes the balancing speed and power losses of ACBN. The stability of the closed-loop system is proved using Lyapunov theory. The cells’ state of charge (SoC) levels required for NMPC are estimated by a state dependent Kalman filter (SDKF). The control scheme is solved using CasADi toolbox, employing the interior point optimizer (Ipopt) algorithm. Despite the modeling mismatch, sensor noises, and an urban dynamometer driving schedule (UDDS) input current profile, the SoC difference stays in a legitimate range of 2%. Moreover, it has been shown that for the same controller, there is 78.6% deviation in the balancing time if static and dynamic model parameters (e.g., charging and discharging path resistances, time delays, and time constants) are ignored.