Novel Hybrid Integration of Zeta Converter and Reinforcement Learning for State-of-Charge Balancing Control in an Electric Vehicle Application

State-of-charge (SoC) balancing control is essential in a Battery management system (BMS) of an Electric vehicle (EV) since it aims to maximize the accessible SoC of each cell, which in turn enhances the overall capacity of the battery system. Cell imbalance can have a negative impact on the battery system, without SoC balancing control, some cells might suffer overcharge or deeply discharge than others, affecting the overall performance of an EV. This work presents a comparative study of three emerging DC-DC converters, notably Zeta, SEPIC, and Ćuk converters as well as three controllers namely Proportional integral (PI), Artificial neural network (ANN) and Reinforcement learning (RL) to select the best converter and controller. The comparative study demonstrated that a Zeta converter with an RL controller is the most efficient in terms of output voltage ripple, voltage stress on output voltage, and settling time. A simulation model is developed in MATLAB/Simulink using twenty Lithium-ion battery (Li-ion) cells where this integration intelligently selects active cell combinations to meet the load, aiming to perform rotation among the cells so that they are not overcharged or deepl