Performance Analysis of Model Predictive Controllers for Active Cell Balancing of Electric Vehicle Battery

One of the functionalities of a battery management system (BMS) in an Electric Vehicle (EV) is to ensure that charge among cells is equally distributed, potentially leading to enhanced performance and reliability. In this context, Model Predictive Control (MPC) has emerged as a suitable candidate. In this work, a battery pack consisting of two serially connected cells along with the bi-directional buck-boost converter is employed to simulate the balancing phenomena. Moreover, to calculate the mean balancing currents and power losses of active cell balancing architecture (ACBN), a high-fidelity model is considered. The purpose is to rigorously investigate the impact of selecting different cost functions in MPC formulation on the closed-loop performance of ACBN in terms of balancing time <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$t_{b}$</tex> and average power losses <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\bar{P}_{L}$</tex>. To this end, a nonlinear model predictive control (NMPC) is formulated to perform active balancing, and the problem is subsequently solved in CasADi framework employing interior point optimizer (Ipopt) algorithm. To mimic the realistic conditions, a US06 City current profile is taken as in input and the simulation's results demonstrates a noteworthy 38.9 % disparity in <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$t_{b}$</tex> and a 4.5 % difference in <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\bar{P}_{L}$</tex> across distinct cost functions, underscoring the need for selecting the optimal cost function within the active cell balancing domain.