提出一个改进的Hopfield神经网络模型,其连接矩阵的对角元T_n=1,同时根据大量的计算机模拟实验并从神经网络的系统能量出发,计算和分析T_n=1及T_n=0两种Hopfield神经网络的各存储样本所处能量状态的分布情况,及其与各存储样本的稳定性和纠错能力的关系,指出改步的Hopfield神经网络其容量和纠错能力均比T_n=0的Hopfield神经网络强。 An improved Hopfield neural network model is proposed, in which the diagonal element of the connection matrix is ​​T_n = 1. At the same time, based on a large number of computer simulation experiments and from the system energy of the neural network, Hopfield neural networks with T_n = 1 and T_n = 0 are calculated and analyzed The distribution of energy state of each stored sample in the network and its relationship with the stability and error correction ability of each stored sample indicate that the modified Hopfield neural network has better capacity and error correction ability than Hopfield neural network with T_n = 0 Strong network.