基于广义回归神经网络的货运量预测

Freight Volume Forecast Based on GRNN

根据货运量形成的原因 ,分析了货运量和相关影响因素之间的关系以及货运量预测的特点。在此基础上 ,建立货运量预测的广义回归神经网络 (GeneralRegressionNeuralNetwork ,GRNN)模型 ,并以我国 1981～ 2 0 0 1年的货运量和相关经济指标的历史统计数据作为学习样本 ,通过拟合训练和外推预测分析 。

According to formation of the freight volume, the paper analyzes the relationship between it and relative factors and the characteristics of freight volume forecast. The paper establishes the general regression neural network (GRNN) model for freight volume forecast. Adapting training and extrapolation forecast are made with specimens out of the historical statistical data of the freight volumes and other economic indices from 1981 to 2001. The paper demonstrates the effectiveness of using GRNN to forecast freight volumes.