| It has become an inevitable trend that new energy electric vehicles(EVs)will replace traditional fuel vehicles due to global energy and environmental crisis.However,the cost and safety of power battery still remains to be the greatest problems that restrict the development and application of EV.Battery management system(BMS),as the link of power battery and EV,is the important guarantee for safe and efficient utilizing of battery,and prolonging battery life.Thus,BMS plays an important role in electronic control components of the whole vehicle.BMS based on distributed architecture has most extensive application in EVs that equipped with high-capacity battery packs.However,this architecture needs many slave controllers and a high performance master,its communication network is more complex and the cost is higher.Besides,tasks and resources allocation between master and slaves is not reasonable.Along with state estimation functions become more and more complex,BMS with traditional distributed architecture exposes many problems that need to be improved.Softwares,such as state estimation algorithm,battery balancing strategy are the core technology of BMS.Coulomb counting and open-circuit voltage calibration are the most applied methods in early stage of BMS products.These open-loop methods have gradually lagged behind the development of BMS.Black-box model methods represented by neural network and machine learning need massive computation,therefore seems not to be a suitable choice in micro-controller applications.Methods based on state-space model have the most extensive application so far.How to identify battery model parameters and improve the accuracy of SOC estimation in this method are the key problems to be solved urgently.Besides,hundreds of battery cells in pack have inconsistencies on parameters such as internal resistance and capacity during the process of manufacturing and utilizing.Improvements of these consistencies are the key points to improve energy utilization efficiency and prolong mileage range of EV.In order to solve the problems above,main content of this paper includes the following aspects:(1)Such problems exist in traditional distributed BMS as uneven distribution of hardware resources and high cost of micro-controller were analyzed.Proposed the architecture based on isolated serial communication bus,and made new integration and distribution for hardware resources,improved the efficiency of resource utilization.(2)Investigated and compared the hardware solutions offered by different semiconductor manufacturers,designed and implemented BMS experimental platform from the aspects of controller,battery pack monitor,balancing driver and topology etc.(3)Transplanted the embedded real time operating system SYSBIOS for BMS hardware platform,made system tasks division and priority allocation,designed synchronization and communication between different tasks.Besides,the operation of the whole embedded system was conducted and tested.(4)Offline calibration and online identification for battery model parameters were compared,and adopted damped recursive least square algorithm to identify model parameters and combined extended Calman filter algorithm to estimate state of charge.In addition,robustness of the proposed algorithm in actual working conditions and its adaptability against battery types and aging were tested.(5)The accuracy of data acquisition module in BMS was verified through massive experiments.On the basis of these verifications,a series of battery equalization experiments were conducted.In these experiments,the feasibility and effectiveness of the proposed balancing topology and strategies were verified through static and dynamic equalization. |
PDF Full Text Request