Abstarct: In this thesis, the problem of balancing rechargeable batteries with a focus on lithium-polymer batteries is addressed. In order to achieve an effective and practical balancing system, it is essential to decompose the main problem into several sub-problems. For a balancing system, estimating the battery state of charge is of particular importance, necessitating the accurate modeling of a battery. In this thesis, alongside proposing a novel machine learning-baxsed method for cell model identification, the state of charge estimation is performed using various nonlinear observers in both deterministic and stochastic spaces. One of the critical issues, especially among lithium-polymer batteries with high discharge rates, is appropriate charge transfer and rapid balancing. Therefore, an innovative strategy is proposed for direct charge transfer between cells. This strategy, besides increasing speed and efficiency, prevents an increase in size and manufacturing costs. A suitable charger circuit is among the requirements for rechargeable batteries. Moreover, the proper performance of the battery balancing system during charging may be affected by electromagnetic interferences caused by high-frequency switching of the charger converter. Therefore, a new control perspective is proposed to regulate the converter output voltage while simultaneously incorporating soft switching in both turn-on and turn-off states of the switch. To control the converter output voltage, an input-output linearizing controller is used. Upon validating the proposed methods through simulation results using appropriate software, the proposed balancing system for charge transfer and the charger converter are constructed experimentally, and the proposed system is validated under practical considerations. The balancing system consists of 5 series-connected lithium-polymer cells. The output results demonstrate the precise performance of all proposed methods.