第二类Chebyshev前向神经网络权值直接确定及结构自适应确定  

Weights-direct-determination and structure-adaptive determination of feed-forward neural network activated with the 2nd-class Chebyshev orthogonal polynomials

作　　者：肖秀春[1,2] 张雨浓[2] 姜孝华[2] 邹阿金[1,2] 

机构地区：[1]广东海洋大学信息学院,广东湛江524088 [2]中山大学信息科学与技术学院,广州510275

出　　处：《大连海事大学学报》2009年第1期80-84,共5页Journal of Dalian Maritime University

基　　金：国家自然科学基金资助项目(60643004;60775050);中山大学科研启动费、后备重点课题资助项目

摘　　要：为克服BP神经网络模型及其学习算法中的固有缺陷,构造了第二类Chebyshev前向神经网络模型,提出该神经网络模型权值直接确定法和结构自适应确定法.理论分析及仿真实验均表明,该系统弥补了BP神经网络的某些固有缺陷.相比同构型BP神经网络,其计算速度和工作精度均有大幅提高.To remedy the weaknesses of back propagation（BP） neural network model and its learning algorithm, a weights-di- rect-determination and structure-adaptive-determination method of feed-forward neural network activated with the 2nd -class Chebyshev orthogonal polynomials was developed. Theoretical analysis and simulation results both show that the proposed sys- tem can remedy the weaknesses of BP neural network model and its learning algorithm, and the calculation speed and working precision improve a lot comparing with the same structured BP neural network.

关 键 词：神经网络 正交多项式 权值直接确定 网络结构 自适应 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]