基于灰色系统和神经网络的实时交通量预测组合模型研究

Research on Combined Model for Real-time Traffic Volume Prediction Based on Grey System and Neural Network

介绍了将灰色系统理论和人工神经网络模型作为预测工具所具有的优缺点,建立了单一的GM(1,1)灰色系统模型。对传统的BP神经网络模型进行改进,在权值函数中加入一个动量因子作为阻尼系数,可大幅降低其容易陷入局部极小值的可能性。同时对学习率加以改进,使其能进行自我调节,于是构建了单一的BP神经网络模型。将单一的灰色模型和BP网络模型进行有机融合,得到了灰色系统—神经网络的组合模型。为验证这3种模型的预测效果,选取某一高速公路路段的单向交通量数据,通过Matlab软件编程进行拟合和预测,发现所建的组合模型综合预测效果最佳。

In the paper the advantages and disadvantages of grey system theory and artificial neural network are introduced when they are taken as prediction tool,and single GM（1,1）grey system model is established.To improve traditional BP neural network model,adding one momentum factor as damping coefficient in weight function can dramatically reduce the possibility of falling into local minimum.And learning rate is improved to realize its self-regulation,the single neural network model is constructed.The single grey model and BP neural network model have been fused organically to get the combined model of grey system and neural network model.In order to verify the three model＇s prediction effects,unidirectional traffic volume date of an expressway section is selected and these dates are used to fit and predict,and the results show that the combined model＇s prediction effect is the optimal.