基于自适应CPN的规则自学习模糊控制器设计

基于扩展Ｋｏｈｏｎｅｎ的自组织映射和Ｇｒｏｓｓｂｅｒｇ的竞争算法，得到了一种高度自适应的双向对传网络（ＣＰＮ），并给出了一类基于此类网络的模糊控制器．根据经验知识对网络离线学习，确定基本的控制规则，利用在线学习算法，实现了规则的自学习．在此算法中，控制器能够自动确定满足控制要求所需规则的个数，并且根据控制目标能够自动获取和校正控制规则．仿真结果展示了控制器应用具体对象的实时控制效果

Research on Rule Self-Learning Fuzzy Controller Based on Adaptive CPN

By extending Kohonen self organized mapping and Grossberg competitive algorithm into highly adaptive Counter Propagation Network (CPN), a kind of fuzzy controller based on adaptive CPN was researched. At first, basic rules were obtained by off line learning for CPN on experience; then by on line learning algorithm, the number of control rules can be ascertained and a method for generating and tuning the control rules was proposed. A relative rule self learning algorithm was given. Finally, the simulation results show effectiveness and feasibility of the proposed algorithm.