稳定性监控自学习FNN控制器

Real-Time Stable Self-Learning FNN Controller by Means of Genetic Algorithm

针对一类未知的非线性系统，提出一种具有稳定性监控的实时自学习ＦＮＮ控制器．ＦＮＮ控制器采用遗传算法使自学习获得最优的性能指标；实时学习过程的稳定性则由基于李雅普诺夫方法设计的监控器来保证，使得ＦＮＮ控制器的自学习过程总是在全局稳定性的条件下进行．

in this paper, a kind of real-time stable self-learning FNN (Fuzzy Neural Network) controller is proposed which uses genetic algorithm (GA) to search 'optimal' fuzzy rules and membership functions for the unknown controlled plant. Since global stability is a basic requirement for any real-time control system to control an unknown plant, the stability of the control system during the real-time learning stage may be guaranteed by a supervisor which is designed with Lyapunov function. The required performance index of the control system may be achieved by transforming the index function into fitness function during the self-learning process of the FNN controller.The fuzzy controller design approach proposed in this paper combines a priori knowledge of the designer with the learning ability of FNN to achieve optimal fuzzy control for an unknown plant in real-time. The efficiency of this approach is verified by a computer simulation for a nonlinear dynamic system.