论文部分内容阅读
提出了一种自学习模糊逻辑推理网络和自学习模糊控制器的构成方法．这种方法是把ＲＣＥ（Ｒｅｓｔｒｉｃｔｅｄ Ｃｏｕｌｏｍ ｂ Ｅｎｅｒｇｙ）模型进行扩展，使其能够进行模糊逻辑推理，并用于构成基于ＲＣＥ模型的自学习模糊控制器ＲＬＦＣ（ＲＣＥｂａｓｅｄ Ｌｅａｒｎｉｎｇ Ｆｕｚｚｙ Ｃｏｎｔｒｏｌｌｅｒ）．这种方法有以下特点：ａ）学习速度高，追加学习容易；ｂ）网络的信息处理工作单元的个数由自学习决定，通用性好；ｃ）不存在局部极小点问题．自学习模糊控制器ＲＬＦＣ可以直接把熟练者的操作知识转换成模糊控制规则，自动构成模糊控制器．数值仿真实验表明其效果良好．
A self-learning fuzzy logic inference network and a self-learning fuzzy controller are proposed. This method extends the Restricted Coulomb Energy (RCE) model to make fuzzy logical reasoning and is used to construct a RCE-based Learning Fuzzy Controller (RLFC) based on the RCE model. This method has the following characteristics: a) high learning speed, easy to learn additional; b) the number of information processing units of the network by self-learning decision, good generality; c) there is no local minimum problem. Self-learning fuzzy controller RLFC can directly convert the operation knowledge of the skilled person to the fuzzy control rules and automatically form the fuzzy controller. Numerical simulation shows that the effect is good.