为提高公路货运量预测的准确性，依据南宁市历史年份的公路货运量数据建立公路货运量 的BP神经网络预测模型。将模型在MATLAB软件环境下进行编码并运算，通过对数据的反复训 练和学习最终得到预测值。经过实例分析证明基于BP神经网络的货运量预测模型的有效性。为 证明不同方法间的差异性，利用趋势外推法、三次指数平滑法、灰色预测法以及指数回归法对南 宁市公路货运量进行了预测。通过对比分析，得到不同方法的相对误差。可以看出，基于BP神 经网络的货运量预测模型较传统预测方法有较大的优越性，BP神经网络模型能够揭示货运量的 非线性变化关系，准确地拟合原始数据。

In order to forecast the highway freight volume effectively, a forecasting model of highway freight volume was constructed by using the BP neural networks based on the history data of highway freight volume of Nanning. The model was coded and operated under the environment of MATLAB soft⁃ ware, and the predictive value was got by training and studying the data. This forecasting model of freight volume which was based on BP neural network was proven to be effective though the example analysis. In order to prove the differences between different forecasting methods, the trend extrapolation method, three exponential smoothing method, and gray prediction method were used to forecast the freight volume of Nanning. And through comparative analysis, the relative error of different methods can be got. It is observed that the freight volume forecasting model based on BP neural network has a bigger superiority than traditional forecasting method. The BP neural network model can reveal the nonlinear re⁃ lation of freight volume, and fit the raw data accurately.