锂电池SOC估算及改进型DC/DC变换器均衡电路研究

Research on lithium battery SOC estimation and improved DC/DC converter equalization circuit

锂离子电池组的不一致性导致电池组增加过充电或过放电风险,并造成个别电池易于老化,使电池组的可用容量和寿命下降,且过充电产生的热量会影响电动汽车的使用安全性。基于此,提出以开关阵列为核心的双向DC/DC变换器均衡电路,以荷电状态(SOC)极值和均值为均衡开启和结束条件,采用主元分析法(PCA)-蚁群算法(ACO)-Elman动态神经网络-无迹卡尔曼滤波(UKF)联合算法进行SOC估算。实验及仿真结果表明,通过算法间取长补短既克服系统输入变量维数大、Elman神经网络易陷入局部最优、UKF过于依赖模型等缺陷,又使均方差达到1%,提高算法的稳定性,计算速度快,且与其他均衡方案相比,均衡耗时短、效率高。

The inconsistency of the lithium-ion battery pack increases the risk of overcharge or over discharge of the battery pack,and causes individual batteries to be easily aging,which reduces the usable capacity and life of the battery pack,and the heat generated by battery overcharge affects the safety of electric vehicles.Based on this,a bidirectional DC/DC converter equalization circuit with a switch array as the core was proposed,and the SOC(State of Charge)extreme value and average value were taken as the opening and ending condition of equalization.PCA(Principal Component Analysis)-ACO(Ant Colony Optimization)-Elman dynamic neural network-UKF(Unscented Kalman Filter)combined algorithm for SOC estimation.The test and simulation results show that the shortcomings between the algorithms not only overcome the system input variable dimension is large,the Elman neural network is easy to fall into the local optimal,the UKF is too dependent on the model and other defects,but also make the mean square error reach 1%,and improve the stability of the algorithm,the calculation speed is fast,and compared with other balancing schemes,balancing takes less time and has higher efficiency.