应用相空间重构技术,论证了网络流量存在混沌现象,并计算了实际网络流量的关联维,Lyapunov指数;在此基础上对网络流量时间序列建立BP神经网络模型对网络流量进行预测,实例表明,和FARIMA(p,d,q)模型相比,基于混沌理论的BP神经网络模型有较好的预测能力,并具有较长的预测步长。 The phase space reconstruction technique is applied to demonstrate the existence of chaos in the network traffic and to calculate the associated dimension and Lyapunov exponent of the actual network traffic. Based on this, a BP neural network model of network traffic time series is used to predict the network traffic. Compared with the FARIMA (p, d, q) model, the BP neural network model based on chaos theory has better prediction ability and longer prediction step.