一种进化模糊逻辑控制器的新方法

结合进化学习分类器的密歇根和匹兹堡方法的优点，首次将对单条控制规则的评价引入了模糊逻辑控制器（ＦＬＣ）的进化过程中，解决了匹兹堡类型的学习分类器系统“强化信息的带宽窄”的问题，实现了ＦＬＣ在控制器级和规则级的同时进化，控制器的控制规则数目也可以自由变化，实验结果表明新方法有较高的效率，优化的模糊控制器的结构简单，性能良好。

A NEW APPROACH TO THE EVOLVE OF FUZZY LOGIC CONTROLLER

Optimization of FLC (fuzzy logic controller) by genetic algorithm is a promising research area. First, rule based FLC is reviewed briefly in the paper. Then, the merits of Pittsburgh approach and Michigan approach are combined in the new FLC's evolving algorithm, and the method of evaluating individual fuzzy rule is introduced into the evolving procedure. FLC and fuzzy rules can be optimized simultaneously in the algorithm. The number of fuzzy rules is variable together with the fixed length of chromosome in the algorithm. Experiments show that the algorithm is efficiency and the optimized FLC is excellent.