城市短时交通流预测仿真研究

摘要

The paper used grey system and neural network to research city short-time traffic flow forecasting prob-lems. The current traffic flow forecasting methods can not effectively handle the city traffie flow real-time change and highly nonlinear characteristics, which leads to poor prediction accuracy in practical. The paper proposed a combined model based on grey system and neural network to fit and forecast the actual monitoring data and obtain the predictive value and prediction residual. The predictive residual was input to the neural network model for learning, simulation and forecasting. Finally, The sum of the predictive residual value and the forecasting value of the GM( 1,1) was ob-tained as the final prediction result. The model is applied to predict the traffic flow of Guiyang fountain. The experi-ment results demonstrate the accuracy and practicality of the proposed method.%应用灰色系统和神经网络研究城市短时交通流预测问题.针对目前交通流预测方法难以处理城市短时交通流实时变化以及高度非线性特征,导致实际预测精度差的缺陷,提出了一种基于灰色系统和神经网络的组合模型,利用灰色模型对实际监测到的数据进行拟合、预测.得到预测值和预测残差.将预测残差输入到神经网络模型进行残差的学习、仿真和预测,残差预测值和GM(1,1)模型预测值的和值作为最终预测结果.运用组合模型方法对贵阳喷水池路段交通流量进行预测,实验结果证明了组合方法的有效性、可行性.