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 Based on the extended Kohonen self-organizing map and Grossberg’s competition algorithm, a highly adaptive bidirectional adversarial network (CPN) is obtained, and a class of fuzzy controllers based on such networks is given. According to the knowledge of the network off-line learning, to determine the basic control rules, the use of online learning algorithm to achieve the rules of self-learning. In this algorithm, the controller can automatically determine the number of rules needed to meet the control requirements, and control rules can be automatically obtained and corrected according to the control objectives. The simulation results show the real-time control effect of the controller application specific object