基于BP神经网络的SOC估计及铅酸蓄电池特性  

SOC estimation based on BP neural network and characteristics of lead-acid battery

作　　者：赵轩[1] 康留旺[1] 汪贵平[2] 马建[1] 

机构地区：[1]长安大学汽车学院,陕西西安710064 [2]长安大学电子与控制工程学院,陕西西安710064

出　　处：《电源技术》2014年第5期874-877,共4页Chinese Journal of Power Sources

基　　金：陕西省科技计划项目(2010K01-071);中央高校基金基础研究项目(2013G1221027);中央高校基金创新团队项目(2013G3322009)

摘　　要：蓄电池特性是影响电动汽车充电时间和续驶里程的主要因素,针对应用较普遍的铅酸蓄电池展开研究具有重要意义。基于阻容等效电路建立了铅酸蓄电池充放电模型,通过不同倍率的充放电实验获得了模型参数与电池荷电状态(SOC)的关系式。采用BP型神经网络模型对铅酸蓄电池SOC进行估计,在Matlab环境下基于SOC神经网络模型对铅酸蓄电池充放电过程进行仿真研究。仿真结果表明铅酸蓄电池模型可以真实地模拟充放电特性,仿真结果与实验结果的平均误差为2.5%。The battery characteristics are the key factors determining the charging time and driving range of electric vehicle, and it is of great significance to study on the widely used lead-acid battery. The charging and discharging model of the lead-acid battery was built based on the RC equivalent circuit, and the equation between the model parameters and state-of-charge（SOC） was calculated through variable rates＇ charging and discharging experiments.The SOC of the lead-acid battery was estimated with BP neural network model, and the simulation study of the lead-acid battery charging and discharging process were carried out based on the SOC neural network model under Matlab circumstances. The simulation results show that the lead-acid battery model could actually simulate charging and discharging characteristics, and the average error between simulation results and test results could reach2.5%.

关 键 词：铅酸蓄电池 充放电特性 神经网络 SOC 建模仿真 

分 类 号：TM912[电气工程—电力电子与电力传动]