大容量锂电池在线参数辨识及SOC联合估计
|
摘要
为了提高对大容量磷酸铁锂电池的在线联合精度,分别对在线参数辨识、及SOC估计两部分做了研究。对电池建立了二阶RC等效电路模型,求出了状态表达方程式;使用变遗忘因子的递推最小二乘法来进行在线参数辨识。在多脉冲放电实验工况下,离线参数辨识的最大误差为4.86%(0.18V);而采用变遗忘因子递推最小二乘法,在线辨识的最大误差为1.89%(0.07V)。在线参数辨识不仅实现了实时性,也提高了精度。在参数辨识的基础上,分别采用扩展卡尔曼滤波算法(EKF)、无迹卡尔曼滤波算法(UKF)对电池SOC进行联合估计。在多脉冲放电实验工况中,当SOC的初始误差在30%以内时,UKF算法收敛到误差允许范围内的最大时间为400s;EKF算法收敛到误差允许范围内的最大时间为1100s(实验中电池的总运行时间为18000s)。且当SOC初值正确时,UKF的最大误差为3.2%,而EKF的误差约为7.8%。因此,UKF的鲁棒性、精确度明显优于EKF。
In order to improve the on-line accuracy of large-capacity lithium iron phosphate batteries,online parameter identification and SOC estimation were studied separately.A second-order RC equivalent circuit model was established for the battery,a state expression equation was obtained,and a recursive least-squares method for changing the forgetting factor was used to perform on-line parameter identification.In the multi-pulse discharge experimental conditions,the maximum error of off-line parameter identification is 4.86%(0.18 V),while the variable error factor recursive least-squares method,the maximum error on-line identification is 1.89%(0.07 V).On-line parameter identification not only achieves real-time performance,but also improves accuracy.On the basis of parameter identification,the Extended Kalman Filter(EKF) and Unscented Kalman Filter(UKF) were used to jointly estimate the battery SOC.In the multi-pulse discharge experimental conditions,when the initial error of the SOC is within 30%,the maximum time for the UKF algorithm to converge within the allowable error range is 400 s;the maximum time for the EKF algorithm to converge within the allowable error range is 1100 s(in the experiment.The total operating time of the battery is 18000 s).And when the SOC initial value is correct,the UKF′s maximum error is 3.2%,while the EKF′s error is about 7.8%.Therefore,UKF′s robustness and accuracy are significantly better than those of EKF.
In order to improve the on-line accuracy of large-capacity lithium iron phosphate batteries,online parameter identification and SOC estimation were studied separately.A second-order RC equivalent circuit model was established for the battery,a state expression equation was obtained,and a recursive least-squares method for changing the forgetting factor was used to perform on-line parameter identification.In the multi-pulse discharge experimental conditions,the maximum error of off-line parameter identification is 4.86%(0.18 V),while the variable error factor recursive least-squares method,the maximum error on-line identification is 1.89%(0.07 V).On-line parameter identification not only achieves real-time performance,but also improves accuracy.On the basis of parameter identification,the Extended Kalman Filter(EKF) and Unscented Kalman Filter(UKF) were used to jointly estimate the battery SOC.In the multi-pulse discharge experimental conditions,when the initial error of the SOC is within 30%,the maximum time for the UKF algorithm to converge within the allowable error range is 400 s;the maximum time for the EKF algorithm to converge within the allowable error range is 1100 s(in the experiment.The total operating time of the battery is 18000 s).And when the SOC initial value is correct,the UKF′s maximum error is 3.2%,while the EKF′s error is about 7.8%.Therefore,UKF′s robustness and accuracy are significantly better than those of EKF.
引文
[1] 闫金定.锂离子电池发展现状及其前景分析[J].航空学报,2014,35(10):2767-2775.
[2] Shen W X.State of available capacity estimation for lead-acid batteries in electric vehicles using neural network[J].Energy Conversion & Management,2007,48(2):433-442.
[3] Aylor J H,Thieme A,Johnso B W.A battery state-of-charge indicator for electric wheelchairs[J].Industrial Electronics IEEE Transactions on,1992,39(5):398-409.
[4] 雷肖,陈清泉,刘开培,等.电动车电池SOC估计的径向基函数神经网络方法[J].电工技术学报,2008,23(5):81-87.
[5] 方明杰,王群京.基于扩展卡尔曼滤波算法的锂离子电池的SOC估算[J].电工电能新技术,2013,32(2):39-42.
[6] 胡小军.基于无迹卡尔曼滤波的动力锂电池SOC估计与实现[D].中南大学,2014.
[7] 程泽,吕继考,刘继光,等.等效滞回模型在锂离子电池SOC估计中的应用[J].湖南大学学报(自科版),2015,42(4):63-70.
[8] 李超,商安娜.电动汽车用氢镍电池二阶RC模型的研究[J].电源技术,2011,35(2):195-197.
[2] Shen W X.State of available capacity estimation for lead-acid batteries in electric vehicles using neural network[J].Energy Conversion & Management,2007,48(2):433-442.
[3] Aylor J H,Thieme A,Johnso B W.A battery state-of-charge indicator for electric wheelchairs[J].Industrial Electronics IEEE Transactions on,1992,39(5):398-409.
[4] 雷肖,陈清泉,刘开培,等.电动车电池SOC估计的径向基函数神经网络方法[J].电工技术学报,2008,23(5):81-87.
[5] 方明杰,王群京.基于扩展卡尔曼滤波算法的锂离子电池的SOC估算[J].电工电能新技术,2013,32(2):39-42.
[6] 胡小军.基于无迹卡尔曼滤波的动力锂电池SOC估计与实现[D].中南大学,2014.
[7] 程泽,吕继考,刘继光,等.等效滞回模型在锂离子电池SOC估计中的应用[J].湖南大学学报(自科版),2015,42(4):63-70.
[8] 李超,商安娜.电动汽车用氢镍电池二阶RC模型的研究[J].电源技术,2011,35(2):195-197.