基于神经网络辨识的熔融碳酸盐燃料电池（MCFC）建模

针对熔融碳酸盐燃料电池（MCFC）电堆系统过于复杂，难以建模以及已建立的模型过于复杂，难以满足工程上对MCFC系统控制设计特别是实时控制的需要，该文试图绕开MCFC的内部复杂性，提出利用神经网络具有逼近任意复杂非线性函数的能力，将神经网络辨识方法应用到MCFC这种高度非线性系统的建模。以燃料气和氧化剂气体的流速为输入量，MCFC电堆的温度响应为输出量，根据输入输出数据用神经网络辨识建立MCFC电堆系统的温度模型，给出了辨识系统的结构及改进BP算法。仿真结果证明了这种方法的可行性，建立的模型精度较高，它使得设计MCFC的实时控制器成为可能。

Modeling Molten Carbonate Fuel Cell(MCFC) Based on Neural Networks Identification

For the seriously complexes of Molten Carbonate Fuel Cells (MCFC), modeling MCFC is very difficult and the models existed are too complicated to be used as a model for controller design, especially for on-line controlling. In this paper we try to avoid the internal complexes of MCFC and set up the temperature model of MCFC using neural networks identification technology, the flow rate of fuel gas and oxidant gas as the input and the temperature response of MCFC stack as the output. The structure and the novel BP algorithm of neural networks identification system are given. The validity and accuracy of modeling are proved by the Simulation results. The neural networks modeling make it possible to design online controller of MCFC.