一种用于ATM业务量预测的神经网络模型

A Neural Network Model for Traffic Prediction in ATM Networks

提出了一种用于ATM网络业务量智能预测的神经网络模型,并对随机产生的自回归型业务量进行了预测。模拟结果表明,与相关文献给出的神经网络模型相比,该模型不但收敛速度较快,而且推广能力强,提高了预测的精度和可靠性。

A new neural network model,in which active functions are hyperbolictangent and linearity respectively in different neurons of the network,was proposed for traffic prediction in ATM networks.Simulations of the prediction of the autoregressive traffic show that the proposed neural network has a faster convergence speed and better generalization than usual network models.