DHNN优化设计新方法及在板形模式识别的应用

PDF下载HTML

Title:

A novel method of optimal designing DHNN and applied to flatness pattern recognition

文章编号:

1673-4785(2008)01-0250-04

作者:

张秀玲; 张志强;

燕山大学电气工程学院; 河北秦皇岛;

Author(s):

ZHANG Xiu-ling; ZHANG Zhi-qiang

College of Electrical Engineering; Yanshan University; Qinhuangdao 066004; China

关键词:

离散Hopfield神经网络（DHNN）;  随机扰动优化设计;  勒让德多项式;  板形模式

Keywords:

discrete Hopfield neural networks（DHNN）;  random destabilization optimal designing;  Legendre orthodoxy polynomials;  flatness pattern

分类号:

TP18

文献标志码:

A

摘要:

基于离散Hopfield神经网络(DHNN)的联想记忆能力,提出了随机扰动优化设计DHNN的新方法.该方法降低了DHNN对权值矩阵的苛刻要求,避免进入伪稳定点;并将其用于板形模式识别,采用勒让德多项式表示常见的6种板形基模式,不需大量的测试样本来训练网络,是一种更简单、实用的板形模式识别新方法,为实现板形控制提供依据,仿真结果证明了这种方法的可行性

Abstract:

A novelmethod of op timal designingDHNN is p roposed based on random destabilization and its associa2 tive memory. Thismethod reduces the harshness requirement of the weightmatrix and avoids getting into the p seudo stability point. The method is app lied to flatness pattern recognition. Denote seven kinds of basis flatness modes thatmeet with usually by Legendre orthodoxy polynomials, don’t need more samp le in training the network. It is a more simp le and availabilitymethod in flatness pattern recognition and makes reference for carrying out the flatness control. Simulation result shows thisway is p racticable.