利用人工神经网络预测电池SOC的研究  

Research on forecast the state of charge of battery based on artificial neural network

作　　者：齐智[1] 吴锋[1] 陈实[1] 于卿[1] 王国庆[1] 

机构地区：[1]北京理工大学化工与环境学院,北京100081

出　　处：《电源技术》2005年第5期325-328,共4页Chinese Journal of Power Sources

基　　金：国家重点基础研究发展计划资助(2002CB211800)

摘　　要：能源和环境与人类社会的生存和发展密切相关,能源短缺和环保的要求推动了混合动力车的发展。电池是混合动力车的动力源之一,准确和可靠地获得电池荷电状态(Stateofcharge,SOC)是电池管理系统的主要任务之一。在总结现有SOC预测方法的基础上,尝试使用人工神经网络的方法预测动力电池SOC。结果表明人工神经网络对多变量的非线性系统非常有效。另外,不同的数据样本对训练效果存在影响,试验表明用处理后的数据进行预测的结果明显较好。Energy sources and environment are very important in social development. They accelerate the progress of hybrid electric vehicles (HEV). Batteries are one of power sources of HEV. It is the most basic and important task that obtain the state of charge (SOC) of the batteries accurately and reliably in the battery management system. In this article, we summarize the forecast SOC methods in common use, and then try to utilize Artificial Neural Network (ANN) to do this job. We can draw the conclusion that ANN is a proper method for SOC forecast, and results are related to the data processing.

关 键 词：混合动力车 电池 荷电状态(SOC) 人工神经网络(ANN) 多元函数预测 

分 类 号：TM911[电气工程—电力电子与电力传动]