Abstract: Aiming at the shortcomings of static feed forward network and Elman network in network traffic prediction，new modified Elman neural network is proposed，and a learning algorithm based on seasonal periodicity is proposed.With a large amount of network traffic data from the actual network，on basic of which，the network traffic is predicted.Simulation experimental results show this model has better effect of prediction.Compared with traditional linear model，BP neural network model and Elman neural network model，it has higher precision and better adaptability.The result shows the model is feasible，reasonable and effective.

Key words: modified Elman neural network, network traffic, modeling, adaptive boundary value, seasonal periodicity

摘要： 针对静态前馈网络和Elman网络在网络流量预测中的不足，建立了一个基于改进Elman神经网络的流量模型，并提出了一种基于季节周期性学习方法，根据实际网络中测量得到的网络流量数据，对网络流量进行预测。实验结果表明，该模型具有良好的预测效果，相对于传统线性模型、BP神经网络模型及标准Elman神经网络模型具有更高的预测精度和更好的自适应性，应用于网络流量预测是可行、有效的。

关键词: 改进的Elman神经网络, 网络流量, 建模, 自适应边界值, 季节周期性