基于神经网络模型的梯次利用锂电池容量估计  

Capacity estimation of second use of lithium-ion batteries based on a neural network model

作　　者：王玉坤[1] 姜久春[1] 张彩萍[1] 张维戈[1] 马泽宇[1] 

机构地区：[1]北京交通大学电气工程学院,北京100044

出　　处：《电源技术》2014年第4期632-635,共4页Chinese Journal of Power Sources

基　　金：国家"863"计划资助项目(2011AA05A108)

摘　　要：针对梯次利用电池容量一致性较差,传统的容量测试方法耗时、耗力、耗材,通过一整车梯次利用电池容量内阻特性研究,发现容量和内阻没有线性相关性。基于不同电池不同采样时间直流内阻所建立的极化内阻均不相同,建立了不同采样时间直流内阻-容量的非线性神经网络仿真模型,并取30个电池模块不同采样时间直流内阻输入进行了仿真,实现了对梯次利用电池容量的快速估计。与传统的容量测试方法结果对比,仿真结果与测试结果基本吻合,验证了该方法的正确性。The capacities of second use of batteries are poor and inconsistence. The traditional capacity test method is a waste of time, labor and cost. Through the test of the capacities and resistances of the second used batteries eliminated from a vehicle, it is found that their capacity and internal resistance have no linear correlation. The polarization resistances of different batteries and different sampling times were not the same, thus a non-linear neural network simulation model of battery capacity was established, and the model＇s inputs were the internal resistances. The internal resistances of 30 batteries were used for simulation, and the fast estimation of the capacity of second used battedes was realized. By comparing with the experimental results and the simulated results, they coincided with each other, which verified the correctness of the neural network model.

关 键 词：动力锂电池 容量 快速估算 BP神经网络模型 

分 类 号：TM912[电气工程—电力电子与电力传动]