基于MIV的BP神经网络磷酸铁锂电池寿命预测  

Cycle life prediction of LiFePO_4 Li-ion battery based on MIV Algorithm and BP Neural Network

作　　者：张金国[1] 王小君[1] 朱洁[2] 迟忠君[2] 

机构地区：[1]北京交通大学电气工程学院,北京100044 [2]北京电力公司北京电力科学研究院,北京100075

出　　处：《电源技术》2016年第1期50-52,共3页Chinese Journal of Power Sources

基　　金：国家电网公司科技项目(GWKJ201203)

摘　　要：针对锂离子电池循环寿命衰减问题,为了能更加准确地对锂离子电池的循环寿命进行预测,对磷酸铁锂电池全生命周期进行循环充放电测试,获得其相关性能参数,提出基于BP神经网络分析方法建立寿命预测模型。在预测模型基础上,运用平均影响值(MIV)算法筛选模型的输入参数。结果表明,所建立的电池循环寿命预测模型具有较高的精度,符合电池的实际运行特性,对解决电池寿命评估周期长和成本高等问题具有重要意义。In order to deal with the declining of cycle life of lithium batteries and to predict their cycle life more accurately, a circulation test of charging and discharging during the whole lifecycle of lithium iron phosphate batteries was conducted and related performance parameter was got. Then a life prediction model was proposed based on the analytical method of BP network analysis. Based on the prediction model, the algorithm of mean influencing value （MIV） was applied to filter the input parameters of the model. The results indicate that the prediction model of cycle life of lithium batteries established possesses a higher precision, which is in accord with the actual running characteristic of the batteries and is of significance to solve the problems of long period and high cost of assessment of battery life.

关 键 词：磷酸铁锂 循环寿命 神经网络 MIV 

分 类 号：TM912[电气工程—电力电子与电力传动]