目前有许多神经网络模型以 Hebb 规则做为学习的基础.在 Hebb 规则中不包含神经元的几何结构信息.但对点数相同的二维点阵和一维点阵,大脑易于记忆二维点阵,因为它有更多的几何结构特征.Kohonen在中论述了生物神经网络的结构特征.本文在Hopfield 网络中引入几何结构信息,对其联想记忆过程做了理论分析和实验研究,试图研究单层神经网络的结构特征对联想记忆的影响,并探讨实现局域连接的可能性.1 Hopfieid 神经网络模型及结构信息的引入Hopfield在文中提出了一个神经网络离散模型.此模型中包括 N 个神经元,它们 At present, many neural network models are based on the Hebb rule, which does not contain information about the geometric structure of neurons, but for two-dimensional lattice and one-dimensional lattice with the same number of points, the brain is easy to remember two-dimensional lattice , Because it has more geometric structural features.Kohonen discusses the structural characteristics of biological neural networks.In this paper, the geometry information is introduced into the Hopfield network, and the theoretical memory and experimental study of its associative memory are made in this paper. Neural network structural features on associative memory and explore the possibility of local connectivity.1 Hopfieid neural network model and the introduction of structural information Hopfield proposed in this paper a neural network discrete model.This model includes N neurons ,they