采用遗传算法直接获取模糊控制器的查询表

在实际应用中，模糊控制器常常被转化为一个查询表，以减小实时运行中的计算量。这样的查询表仍然能够保留原模糊控制器的性能特征，但另一方面这个查询表是如何获得的对控制器来说并不是本质的，控制器的性能只取决于查询表本身。这样，在模糊控制器的结构下采用其他方法直接生成这种查询表，或许能够比采用模糊逻辑的方法更为有效。本文给出了采用遗传算法直接搜索查询表的方法，以获取性能优良的控制器，给出了查询表式下二队控制

Direct Derivation of Fuzzy Controller Look-Up Tables Using Genetic Algorithms

In physical implementation, a fuzzy controller is usually reduced to a look up table in order to minimize real time computational complexity. Most features of the original fuzzy controller is still maintained by the look up table. While it is this look up table which is essential to the performance of the controller, the way in which this table is obtained is of no effect. This motivated the attempt of deriving a look up table directly by other means, which may be more effective than using fuzzy logic. A second order controller represented by a look up table is formalized and genetic algorithms are used to find suboptimal controllers. A genetic representation of look up tables is given, which greatly eliminates the information redundancy in such tables by their symmetry and monotonicity. It is shown by a simulation example that using the given method a controller can easily be obtained with its performance better than that of another controller based on fuzzy control.