基于模糊—神经融合的自适应模糊控制系统研究

模糊逻辑系统与人工神经网络各具优势，前者善于利用专家语言信息，后者有强大的学习能力，两者的结合可以取长补短。基于模糊逻辑系统与神经网络技术提出一种自适应模糊控制系统，其特点是模糊控制器具有多层前向网络结构。基于一种近最优的性能指标导出其参数自适应的误差反向传播算法。为了克服传统算法收敛慢的缺点，提出用模糊逻辑来调整学习过程的方法。通过倒立摆平衡控制仿真研究验证了所提出的自适应模糊控制系统及其快速学

A research on adaptive fuzzy control system based on fuzzy-neural fusion

Fuzzy logic system is good at utilizing experts' information, while neural network possesses powerful leaming ability, and they can leam from others' strong points to offset its weakness by combining two methodologies . An adaptive fuzzy control system based on fuzzy logic system and neural network is proposed in this paper. This system features a fuzzy controller with a multi-layer network structure, so. naturally a error back-propagation algorithm can be derived from a near-optimal performance index to adaptively tune the rules' parameters. In order to improve the convergence rate, a fuzzy logic method is used to adjust the learning rate. At last, simulations for controlling invert-pendulum prove the adaptive fuzzy control system and its fast learning algorithm proposed in this paper are effective.