用模糊RBF神经网络简化模型设计多变量自适应模糊控制器

Design a Multivariable Adaptive Fuzzy Controller by Fuzzied Radial Basis Function Network Simplified Model

针对多变量系统实时性要求 ,提出模糊径向基 (RBF)神经网络结构的简化模型及相应算法 ,并对由此简化模型设计的多变量模糊控制器模糊规则的在线自学习算法进行分析 ,提出一种系统动态增益的处理方法和基于过程最优的改进方案 .仿真实验结果表明该控制器可实现实时自适应控制 ,改进算法是有效的 .

This paper presents a simplified model and the corresponding algorithm of fuzzy radial basis function (RBF) networks to solve the real time control of multivariable process. Authors also analyze the self learning algorithm of multivariable fuzzy controller designed by this simplified model and discuss a new method to treat system dynamics gain in the self learning algorithm. Furthermore a modified self learning algorithm is presented based on process parameters optimization. Finally computer simulation results of an industrial process verify that the simplified model and the modified algorithm are available and effective.