基于改进RBFNN的SOFC辨识建模

针对现有的固体氧化物燃料电池(SOFC)模型过于复杂,难以满足控制系统的设计需要的弊端,基于一种改进的径向基函数神经网络(RBFNN)辨识技术建立了SOFC的非线性模型.在建模过程中,以SOFC的燃料利用率为模型的输入,电压和电流为模型输出.利用800组实验数据作为训练样本,建立了SOFC的电流-电压辨识模型.仿真结果表明了所建模型的有效性和精度.该模型的建立为先进的控制策略研究奠定了基础.

Modeling SOFC Based on Improved RBFNN Identification

According to the drawbacks of the existed mathematic models, which are too complicated to meet the design demand of SOFC control system, a nonlinear model based on a kind of improved RBF neural network (RBFNN) identification technique is presented in this paper. The fuel utilization of the SOFC is taken as the input, the voltage and current density as the outputs of the neural network model. With 800 groups of experimental data as the training samples, a cell voltage and current density identification model of the SOFC is established. The simulation results show the validity and accuracy of the model. Furthermore, based on this RBFNN identification model, some advanced control schemes can be developed.