基于组合预测模型的铁路集装箱运量预测

为了克服单项预测模型的单一性和片面性等缺点,本文应用组合预测模型对铁路集装箱运量进行预测,以提高预测的准确性。通过对运量的历史数据分别采用多项式曲线模型和灰色预测模型建立单项预测模型,再利用径向基神经网络对两个单项预测模型的结果进行组合预测。研究结果表明,相比于两单项预测方法,组合预测方法所得运量的相对误差分别提高了3.19%和12.76%。最后,应用马尔科夫链模型对组合预测的结果进行分析和修正,增加预测结果的可靠性。

Forecast of Railway Container Freight Volume by Using a Combinatorial Model

The forecast of railway container freight volume has significant effect on the operation and development of the railway.The existing forecast models can forecast a single index only,which is not accurate enough.To overcome this disadvantage,the combinatorial forecast model is adopted to forecast railway container freight volume.Based on the historical data,individual index forecast models are derived by using linear polynomial and grey models,respectively.Then,the individual index forecast models are combined by using radial basis function(RBF) neural network.Analysis shows that,in comparison with two single index forecast models,the combinatorial forecast model can improve the forecast result of relative error by 3.19% and 12.76%,respectively.Finally,the combinatorial forecast result is analyzed and modified by Markov chain model.