摘要
首先介绍并分析了固体氧化物燃料电池(SOFC)的工作原理和理论电压模型。然后,针对SOFC系统过于复杂,理论电压模型存在明显不足的特点,试图绕开SOFC的内部复杂性,利用广义回归神经网络(GRNN)对SOFC系统进行辨识建模。模型以电池工作温度为神经网络辨识模型的输入量,电池电压/电流密度为输出量,利用750组实验数据作为训练样本,建立了SOFC在不同工作温度下的电池电压/电流密度动态响应模型。仿真结果表明了该方法的有效性,所建模型精度也较高。
In this paper, the working principle of the solid oxide fuel cell （SOFC） and the theoretic voltage model are introduced and analyzed. Then aiming at the serious complexity of SOFC and the apparent shortcomings of the theoretic voltage model, the paper tries to avoid the internal complexity of SOFC and set up a dynamic response model of voltage and current density at different operating temperatures by using generalized regression neural networks （GRNN）. The operating temperature of SOFC is taken as the input and the voltage and current density as the output of the neural networks model. With 750 groups of experimental data as the training samples, a cell voltage and current density identification model of SOFC is given. The simulation results show the validity and accuracy of the model.
出处
《计算机仿真》
CSCD
2007年第2期232-235,共4页
Computer Simulation
基金
国家"863"计划发展基金资助项目(2003AA517020)
关键词
固体氧化物燃料电池
广义回归神经网络
模型
Solid oxide fuel cell （SOFC）
Generalized regression neural networks （GRNN）
Model
