摘要
文中首先对模糊系统的两类主要的学习算法:梯度下降和遗传算法方法进行了深入分析,并指出了存在的问题.然后,在此基础上提出了一种针对半梯形和三角形隶属度函数的保证隶属度函数ε完备性和模糊集语义一致性的参数调整方法.并基于上述方法实现了一种新的基于遗传算法并利用梯度下降的快速模糊系统学习算法.最后通过实例进行了模拟,验证了该方法的高效性。
In this paper, two kinds of learning methods of fuzzy systems are analyzed first,using genetic algorithm and the gradient descent method. They have completeness of membership functions and fuzzy rules and other problems, such as the damage of the shapes of membership functions. The damage of completeness of membership functions leads to no useful rules which are available when some data are inputted. Then, a method that guarantees the ε\|completeness of membership functions and consistence of fuzzy sets semantics is proposed. Moreover, a new fast learning method of fuzzy systems both are based on genetic algorithms and gradient descent method is proposed. Some experiments are also made and the simulation result is presented to show the high effectiveness and some other advantages of guaranteeing the completeness of membership functions and consistence of fuzzy sets semantics.
出处
《计算机研究与发展》
EI
CSCD
北大核心
1999年第9期1080-1085,共6页
Journal of Computer Research and Development
基金
上海市教委重点学科项目基金
关键词
隶属度函数
语义一致性
模糊集合
模糊系统
completeness of membership function, consistence of fuzzy sets semantics, genetic algorithms,gradient descent method