Development Of Battery Management System In Pure Electric Vehicle

Under the background of energy saving and emission reduction, electricvehicle has become a hotspot. Battery management system which is one of thethree core technologies of electric vehicles, monitors the work state of powerbattery pack in real time, estimates the battery’s state of charge, equalizes thenon-uniformitys. It’s important for the battery safety and cycle-life.Firstly, according to the function of BMS, this dissertation focuses on theresearch on development approach of distributed architecture BMS, includingsoftware and hardware design method. In the hardware design, as there isaccumulated potential and the accumulated potential of each cell is different fromthat of another, the local electrical unit is divided into four modules. In thesoftware design, except for realizing the required basic functions, it’s necessaryto improve the portability of software. The hierarchical modular design method isutilized to decouple the basic and application software, which reduces thesoftware dependence on the bottom hardware.Then, for the non-linear, time-varying battery system, a sliding modeobserver for estimating SOC is proposed. An improved Thevenin battery model isachieved for lithium-ion polymer battery of electric vehicle, and the modelparameters are identified online by adopting the extended Kalman filter (EKF)algorithm. By introducing Luenberger-type feedback terms, a sliding modeobserver for estimating SOC is proposed, and a sufficient condition is derived toguarantee the convergence of the observer. The proposed method is verified andevaluated by experiments. The results show that, SOC estimation with the slidingmode observer has high accuracy, and have good convergence.Finally, in order to equalize the non-uniformity of battery packs better, andthe passive equalization has disadvantage of low efficiency and large heat, theactive equalization is studied. The multiple transformers and battery balancer ICare adopted to achieve bidirectional equalization of any cell in the module, whichavoid the issue of control complexity. This solution meets the requirements in theverified and evaluated experiments.