模糊逻辑系统的快速构造与优化

提出一种直接利用均匀分布于待逼近系统输入空间的Ｉ／Ｏ数据，快速构造满足一定精度要求的模糊逻辑系统方法，并从理论上证明了该方法的可行笥。在此基础上采用一种新型的ＧＡ＋ＢＰ混合自治对模糊逻辑系统进行优化，以求用最少的规则数实现满意的精度。数字仿真结果表明这种快速构造和优化方法是可行和高效的。

Fast Construction and Optimization of Fuzzy Logic Systems

A fast construction method of fuzzy logic systems was presented and the optimization of parameters and structures of such systems was discussed. Some input/output data of the system to be approximated, which are uniformly distributed in the input space, were directly used to form the rules and membership functions of the fuzzy logic system. Theoretic analysis shown that the fuzzy logic system constructed using such method could easily approximate any bounded continuous system at any precision. To obtain the given precision using the least rules, a new GA BP hybrid learning algorithm and the dichotomy were used to optimize the parameters. Numeral results shown that the fast construction method and the optimal algorithm are feasible and effective.