基于核函数切换和支持向量回归的交通量短时预测模型

Traffic volume prediction based on support vector regression with switch kernel functions

基于高速公路交通量短时变化的非线性、不确定性和复杂性,利用支持向量回归模型,提出一种核函数切换的预测方法.首先,通过历史数据构建不同核函数的支持向量回归模型并对历史数据进行拟合,根据拟合的误差确定不同时刻对应的最优核函数类别;然后根据历史数据及确定的不同时刻的核函数类别训练支持向量分类机;最后利用支持向量分类机确定预测时刻最优的核函数类别,选取相应的支持向量回归模型进行预测.实例分析表明,与传统的支持向量回归模型相比,含核函数切换的预测方法预测精度较高,且具有较好的鲁棒性.

To simulate the nonlinear,probabilistic and complicated patterns in the short-term change of the highway traffic volume,a prediction model was proposed based on support vector regression and switch kernel functions. First,support vector regression models were built with different kernel functions by the historical data and the best kernel function was obtained using the fitting error.Then,a support vector machine model was trained. Finally,the best kernel function for the prediction interval was selected and the corresponding support vector regression model was implemented. A case study was used to evaluate the performance of the proposed model. The result shows that the model is superior to the traditional support vector regression model on the predicted accuracy,and thus it is more robust.