基于改进的GM(1,1)的长期交通量预测模型

A forecast model for long-term traffic volume based on improved GM(1,1) model

针对传统的GM(1,1)模型在预测高速公路交通量中存在的误差过大的问题,通过对原始数据进行滑动平均处理,减少数据在统计过程中的随机误差和人为误差。利用等维灰数递补预测模型进行交通量预测,在数据列中补充新的数据,去掉老的数据,使模型得到改进。利用改进的新模型去预测下一年的数据比用原模型更加合理,更接近实际。研究结果表明:利用等维灰数递补预测模型预测的预测精度是94.24%,比GM(1,1)残差改进模型提高了1.49%,比传统的GM(1,1)模型精度提高了6.94%。适用于交通量的长期预测。

In order to minish the excessive error from traditional GM （ 1,1 ） model in expressway traffic volume prediction, this paper aims to reduce random error and personal error of data in the statistical process by a sliding average processing of original data. The same dimension gray recurrence dynamic model was adopted to predict the traffic volume, and the model was improved by adding new data continuously and removing old data. Using improved data to forecast next years data is more rational and realistic than using the source model. The research results show that, the prediction accuracy of traffic volume forecast with progressive model of equal dimension grey member GM（ 1,1 ） is 99.24%, which improves 1.49% in comparison with residuals improved GM （1,1） model and 6.94% compared to GM （ 1,1 ） model. The progressive model can be used to forecast long - term traffic volume.