基于支持向量回归机的公路货运量预测模型*

为了提高公路货运量预测的能力,应用基于结构风险最小化准则的标准支持向量回归机方法来研究公路货运量预测问题.在选择适当的参数和核函数的基础上,通过对成都公路货运量时间序列进行预测,并与人工神经网络、线性回归分析等方法进行了对比,发现该方法能获得最小的训练相对误差和测试相对误差.

Model of highway freight traffic forecasts based on support vector regression

To improve the forecast ability of highway freight traffic,SVR based on structural risk minimization was applied to forecasting highway freight traffic. By selecting appropriate parameters and kernel function, the proposed approach was used for forecasting highway freight traffic of Chengdu city. Compared with artificial neural network (ANN) and linear regression analysis, experimental results show that the training relative error and testing relative error obtained by SVR is lower than that by ANN and linear regression analysis.