摘要
提出了一种基于模糊集合与前馈神经网络结合的神经网络。这种模糊神经网络(FNN)在处理模糊特征时,能较好地反映出输入值和输出值的隶属度关系。应用经典的BP算法对网络进行训练。为了加快网络的学习过程,我们介绍了一种调整权值和阈值的学习方法,并且采用了C语言编制软件。实验结果表明,FNN技术是一种新颖、有效的方法,能促进智能神经网络的发展。
model of fuzzy neural network(FNN)based on fuzzy sets and feed forward neUral network is presented.The FNN can describe a relationship of value of the membership function of input and output data. A cias-'sic BP algorithm is used. In order to accelerate learning process,a method is introduced to adjust weights and thresholds,also a software is programmed in C language. Experiments and results demonstrate that FNN technique is a novel and potetially powerful approach toward intelligent neural network.
出处
《上海航天》
1995年第4期8-12,共5页
Aerospace Shanghai
关键词
模糊集
隶属函数
BP算法
模糊神经网络
Fuzzy sets,Membership function,BP algorithm,Fuzzy neural network