In the last ten years, an increasing number of researches are engaged in studying the design of Battery Management System (BMS) and the estimation of State-of-Charge (SoC) for high-power Lithium batteries. The BMS is designed to enhance battery performance, extend the calendar life and guarantee the safety. The estimation accuracy is one of the key issues for the BMS and the extensive use of EV/HEV in the future.

In this thesis, a set of battery electrochemical characteristics experiments have been conducted for BMS implementations and SoC estimation. A BMS which was consisted of master system and slave system has been designed with real-time high accuracy measurement and high efficiency active cell balancing. Then an adaptive method based on Coulomb counting combined with Exte...[ Read more ]

In the last ten years, an increasing number of researches are engaged in studying the design of Battery Management System (BMS) and the estimation of State-of-Charge (SoC) for high-power Lithium batteries. The BMS is designed to enhance battery performance, extend the calendar life and guarantee the safety. The estimation accuracy is one of the key issues for the BMS and the extensive use of EV/HEV in the future.

In this thesis, a set of battery electrochemical characteristics experiments have been conducted for BMS implementations and SoC estimation. A BMS which was consisted of master system and slave system has been designed with real-time high accuracy measurement and high efficiency active cell balancing. Then an adaptive method based on Coulomb counting combined with Extended Kalman Filter is developed for the SoC estimation. An improved 3-state RC circuit model, in which the renewal of battery is considered, has been developed. The initial model parameters are identified offline by multivariable linear regression (MLR) and adjusted online to cope with the changing dynamics introduced by the changes in the charge-discharge curves, Coulombic efficiency, and electrochemical phenomena of hysteresis and polarization typically encountered as the characteristics of the LiFePO₄ batteries.

The BMS serves to be a platform with monitoring, balancing and control functions. The experimental tests have shown that the proposed method is practical and can effectively estimate SoC under 5% estimation error, even with noise, miss-matched initial SoC and polarization effect.