正则模糊神经网络对一类连续模糊值函数的普遍近似性  

Universal Approximation of a Class of Continuous Fuzzy Valued Functions by Regular Fuzzy Neural Networks

作　　者：刘普寅[1] 王华兴[1] 

机构地区：[1]国防科技大学系统工程与数学系

出　　处：《电子学报》1997年第11期41-45,共5页Acta Electronica Sinica

基　　金：国防预研基金

摘　　要：本文讨论了正则模糊神经网络对于定义于区间［０，Ｔ０］上连续可减模糊值函数的普遍近似性，在此基础上，证明了［０，Ｔ０］上取值为三角形模糊数的连续递增函数可用正则模糊神经网络逼近到任意精度．In this paper, it is argured that regular fuzzy neural networks can be universal approximators to continuously subtractable fuzzy valued functions, defined on the interval [0, T0]. Then arbitrarily accurate approximations of triangle fuzzy number valued functions that are continuously increasin on [0, T0] by regularfuzzy neural networks are showed.

关 键 词：正则模糊 神经网络 可减模糊值函数 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]