基于ROLS算法的RBF神经网络燃料电池电特性建模  

Electric-characteristic Modeling of a Fuel Cell Based on ROLS Algorithm and RBF (Radial Based Function) Neural Network Identification Technique

作　　者：苗青[1] 曹广益[1] 朱新坚[1] 

机构地区：[1]上海交通大学燃料电池研究所,上海200030

出　　处：《热能动力工程》2005年第4期387-389,共3页Journal of Engineering for Thermal Energy and Power

基　　金：国家863计划基金资助项目(2003AA517020)

摘　　要：提出了一种基于ROLS算法的RBF神经网络辨识建立直接甲醇燃料电池(DMFC)电特性模型的新方法。以电池的工作温度为输入量,电池的电压/电流密度为输出量,利用1200组实验数据作为训练和测试样本,建立了在不同工作温度下,电池的电压/电流密度动态响应模型。仿真结果表明采用RBF神经网络辨识建模的方法是有效的,建立的模型精度较高。An innovative method is presented for the electric-characteristic modeling of a direct methanol fuel cell (DMFC) through the use of ROLS algorithm-based RBF (radial based function) neural network identification technique. With the operating temperature of the cell serving as an input and the voltage/electric current density of the cell serving as an output 1200 groups of experimental data were utilized as training and test samples to set up under various operating temperatures a dynamic response model of the cell voltage/electric current density. Simulation results indicate that the modeling method by using the RBF neural network identification technique is effective with the established model featuring a relative high precision.

关 键 词：直接甲醇燃料电池 RBF神经网络辨识 ROLS算法 

分 类 号：TM911[电气工程—电力电子与电力传动] TP183[自动化与计算机技术—控制理论与控制工程]