基于RBFNN的DMFC温度建模与神经模糊控制研究

为了提高燃料电池的发电性能，直接甲醇燃料电池（DMFC）堆的运行温度应该控制在一个合适的范围内。简单介绍了利用RBF神经网络基于实验的输入输出数据建立DMFC电堆温度模型的方法，避开了电堆的内部复杂性；在控制过程中，将训练好的网络模型作为DMFC控制系统的参考模型，采用一种改进的模糊遗传算法（FGA）在残对神经模糊控制器的参数进行自适应调整。采用最近邻聚类算法小（NNCA）对控制器的模糊规则库进行更新。在仿真实验中，将所提出的算法与非线性PID和传统模糊算法进行比较，结果表明所设计的神经模糊控制器具有较好的性能。

Temperature Modeling of DMFC Based on RBFNN and Neural Fuzzy Control Study

To improve the performance of fuel cells,the operating temperature of Direct Methanol Fuel Cell(DMFC)stack should be controlled within a specified range.Based on the input-output sampled data,the method of modeling DMFC stack temperature with a Radial Basis Function Neural Network(RBFNN)was simply introduced,which avoids the internal complexity of DMFC stack.In control process,with the network model trained well as the reference model of the control system of DMFC stack,a novel Fuzzy Genetic Algorithm(FGA)was used to regulate the parameters of a neural fuzzy controller,and a Nearest Neighbor Clustering Algorithms(NNCA)was adopted to refresh the fuzzy rules.In the simulation,compared with the nonlinear PID and traditional fuzzy algorithms,the neural fuzzy controller designed gets better performance demonstrated by the simulation results.