摘要
一个运行良好的模糊系统,不仅应该具备最优的响应性能,还应该具备最简约的模糊划分和良好的完备性、清晰性,但是目前对于后者的研究甚少。因此,该文提出了最优模糊划分的定义,它包括两个部分:最简约模糊划分和模糊划分的完备-清晰性。然后将这两项指标加入到进化策略的适应值函数中,用于对模糊系统的寻优。仿真试验表明,即使在先验知识较少的情况下,该方法依然可以设计出具有较好系统响应性能的模糊系统,并且模糊划分个数最简,完备-清晰性较好。
A good fuzzy system should exhibit not only good approximation, but also minimal fuzzy partition, good completeness and distinguishability. But fewer efforts have been done for the latter. So a definition for optimum fuzzy partition is proposed in this paper. Its main aspects include the minimal form and completeness-distinguishability of fuzzy partition. These conditions are added to the fitness function of the evolutionary strategy for optimizing the fuzzy system. To check the effectiveness of the suggested approach, an example for nonlinear plant control is examined. It is verified that the fuzzy system works very well, especially when the priori Knowledge are insufficient.
出处
《中国电机工程学报》
EI
CSCD
北大核心
2004年第9期147-152,共6页
Proceedings of the CSEE
关键词
进化策略
最优模糊划分
模糊系统
适应值函数
Fuzzy System
Evolution Strategy
Optimum Fuzzy Partition
Minimal Fuzzy Partition
Completeness-Distinguishability
