基于LS-SVM的铅酸蓄电池极化电势预测模型研究  

Research on estimating polarization potential of lead-acid battery based on LS-SVM algorithm

作　　者：李匡成[1] 杨亚丽[1,2] 孙磊[1,3] 刘瑾 

机构地区：[1]装甲兵工程学院控制工程系 [2]96111部队 [3]73089部队 [4]宁夏军区自动化站

出　　处：《电源技术》2014年第4期724-727,共4页Chinese Journal of Power Sources

摘　　要：蓄电池充电过程中,极化电势的累积阻碍了充电电流的接受能力,降低了快速充电的效率。针对极化电势难预测的问题,提出了最小二乘支持回归机预测极化电势的方法。确定了模型的输入量;经遗传算法参数寻优,确定了模型中的参数。仿真结果证明,采用LS-SVM算法预测极化电势,在小样本的条件下能达到预期的辨识精度,且泛化能力强。Polarization potential impedes the acceptability of charging current, decreasing the charging efficiency. Due to the test difficulty of polarization potential, an algorithm was presented to estimate it. Determining the inputs of the model, the parameters were defined based on genetic algorithm. Though simulation experiment, it is proved that the model of LS-SVM is fully accurate with stronger generalized capability on the condition of small-scale dataset.

关 键 词：铅酸蓄电池 极化电势 最小二乘支持向量机 遗传算法 

分 类 号：TM910.6[电气工程—电力电子与电力传动]