基于T-S模型的扩展型模糊神经网络及应用  

Extended fuzzy neural network based on T-S model and its application

作　　者：韩敏[1] 范迎南[1] 

机构地区：[1]大连理工大学电子与信息学院,辽宁大连116023

出　　处：《系统工程学报》2007年第5期532-538,共7页Journal of Systems Engineering

基　　金：国家自然科学基金资助项目(60674073);国家重点基础研究发展计划(973)项目(2006CB403405);国家科技支撑计划资助项目(2006BAB14B05)

摘　　要：针对基于T-S模型的模糊神经网络的局部逼近缺陷,提出了一种基于T-S模型的扩展型模糊神经网络,从训练样本特性和网络结构两个方面来提高网络模型的泛化能力.利用先验知识和模糊推理的方法对样本集进行分析和分类处理,使样本集更加规范;并采用模糊规则推理动态调整正则项系数的方法来减小网络结构.仿真结果表明,所提出的网络具有更快的收敛速度和良好的泛化能力.In order to overcome the drawback local approximation of the fuzzy neural network based on T-S model, an extended fuzzy neural network based on T-S model （EFNN-TS） is proposed in this paper. Two aspects of the characteristics of training samples and network structure are considered to enhance the generalization ability of the network. For normalizing the pattern set, the training patterns are classified and processed by using prior information of the patterns and fuzzy inference approach. Regularization is added whose coefficient can be adjusted dynamically by fuzzy rea- soning to simplify the structure of the network. The simulation results indicates that the proposed method has more rapid convergence and better generalization ability.

关 键 词：模糊神经网络 投标报价 泛化能力 正则项系数 样本分类 

分 类 号：TP118[自动化与计算机技术—控制理论与控制工程]