Abstract: The network traffic is the important parameter that measures the burden of network movement and network appearance.It also plays an important role in network layout，traffic anagement.In traffic management，traffic model is used to evaluate the mechanism of join control and predict network performance.The grey model and neural network have good effect in reflecting the variable trend of data.With the development of grey neural network theory and its widely used，many improved and new generation methods have been proposed.On the research of neural network，this paper added a compensated error.so the prediction value equals to the output value of grey neural network model plus the compensated error signal.The simulation resuluts show that the integrated model can improve the prediction precision obviously compared to the other algorithm.

Key words: network traffic, grey model, neural network, grey neural network, neural network compensator, prediction

摘要： 网络流量是衡量网络运行负荷和状态的重要参数，也是网络规划、流量管理等方面起着重要作用的重要参数。在流量管理中，流量模型用于评价接入控制机制和预测网络性能。灰色模型和神经网络在反映数据的趋势性变化上效果明显，随着灰色神经网络的发展及其广泛应用，越来越多的方法已经被提出。文中利用神经网络补偿器获得误差补偿信号，则最终的预测值为灰色神经网络模型的预测值加上误差补偿。仿真结果验证了所提方法的有效性。

关键词: 网络流量, 灰色模型, 神经网络, 灰色神经网络, 神经网络补偿器, 预测