Abstract: Combined forecast method compared with individual forecast methods can improve the precision and stability of forecast results，and it is widely used in practice．First，the basic idea of combined forecast is summarized．Then，combined-forecast model based on BP neural network is introduced and a calculation example of highway cargo volume in Jilin Province is presented．The forecasting result is very ideal and the applicability of this combined model is proved by mathematical statistics，the feasibility and applicability of this combined model are proved by statistical analysis and practice．This model for forecasting highway transportation volume is effective and feasible．

Key words: combination forecast, highway transportation volume, BP neural network, model

摘要： 组合预测方法与单一预测方法相比可以提高预测的精度和稳定性，因此得到广泛的应用。本文首先概述了组合预测的基本思想，然后介绍了基于BP神经网络的组合预测模型，并以吉林省公路货运量为例给出计算实例，实例的预测结果非常理想，同时也用数理统计的方法证明了此模型的适用性。统计分析和实践都证明此模型的可行性和适用性，说明将此模型用于公路交通运输量预测是有效可行的。

关键词: 组合预测, 公路运输量, BP神经网络, 模型