摘要
通过对已有的一些基于遗传算法优化模糊系统方法的分析,指出了它们存在的一些缺陷。提出了一种新颖的基于优化模糊(GA)GA推理神经网络的方法,并给出了相应的优化算法。这种方法可以对模糊推理系统中的所有结构和参数同时或分别进行优化。在此基础上,还讨论了模糊推理神经网络的精简问题,如无用模糊规则的删除。最后通过实例验证了该方法是一种很有效的方法。具有易理解、精度高、收敛快、泛化能力好且能全局收敛的优点。
By analyzing certain methods for optimization of fuzzy system based on genetic algorithm (GA), the author points out some of their drawbacks, proposes a novel method for optimization of fuzzy inference neural network based on GA, and offers a corresponding optimization algorithm. This method can optimize all parameters and structures of the fuzzy system simultaneously or respectively. Then, the author discusses the problem of reducing the redundancy of fuzzy inference neural network such as the cutting off useless fuzzy rules and combination of a fuzzy rule with the same consequent. Finally, the author testifies the method by example simulation, which proves to be valid and which has the advantages of comprehensibility, high precision, good generalization, rapid and overall convergence. ;;;;
出处
《计算机工程》
CAS
CSCD
北大核心
2002年第7期23-25,121,共4页
Computer Engineering
基金
铁道部基金资助项目
关键词
GA
优化
模糊推理
神经网络
遗传算法
模糊规则
隶属度函数
Genetic algorithmFuzzy inference neural networkOptimizationFuzzy rulesMembership functions
