一类新的模糊控制算法

A new class of fuzzy control algorithms

本文针对模糊控制器输入输出是确定量的特点,提出一类新的简单的推理决策算法。首先对隶属函数进行一般化的定义,然后用输入模式和规则模式表示过程测量值和用语言描述的人的控制规则。推理算法简化为在输入空间中两类模式的匹配度计算和最后的加权平均计算。在多维体积比和相对距离两种方法度量匹配度情况下,给出了4个具体算式。及算法的完整步骤,最后讨论了它的优点和局限。

Noting that the inputs and outputs of the Fuzzy Controller are non- fuzzy, this paper presents a new and simple class of reasoning and decision algorithms. First the membership functions are generally defined. Then the actual inputs and control rules are expressed by the input pattern and the rule patterns respectively. The control algorithms are reduced to the match degree computations of two patterns in the input space and a final weighted average calculation. Two methods of measuring match degree are proposed and four specific algorithms are presented. Finally the complete computing steps are given and some advantages are discussed.