摘要
介绍了一种基于动态模糊神经控制器的控制系统,通过在模糊神经网络控制器的第2层引入动态递归环节,使其具有动态映射能力,并提出了动态模糊神经控制器的混合学习算法,即先采用免疫遗传算法的“粗”学习,再采用BP梯度算法的“细”学习.通过对锅炉主汽温控制的仿真表明了该网络结构和训练方法是可行的和有效的.
A novel control system based on the dynamical neuro-fuzzy network is discussed. The network has the ability of dynamical mapping by adding recurrent nodes in the second layer of the function network of the network. The hybrid training algorithms of the dynamical neuro-fuzzy network is proposed as well, where the "thick" learning of the immune genetic algorithm is adopted firstly, then the "thin" learning of BP algorithm is used. The simulation on main steam temperature control system shows that the network structure and the training algorithms are feasible and effective.
出处
《上海电力学院学报》
CAS
2005年第4期310-315,338,共7页
Journal of Shanghai University of Electric Power
基金
上海市教委发展基金(020K02)
关键词
动态模糊神经控制器
免疫遗传算法
主蒸汽温度
dynamical fuzzy-neural controller
immune genetic algorithm
main steam temperature