基于Gaussian-支持向量回归机的高速公路短时交通量预测

Prediction of Freeway Short-term Traffic Flow Base on Gaussian-SVR

针对高速公路短时交通量的实时性、波动性和非线性的特点,将支持向量回归机(SVR)应用于高速公路短时交通量预测,并采用Gaussian损失函数来代替ε-不敏感损失函数,对原始序列进行降噪处理,为了更好的优选SVR模型参数,采用遗传算法(GA)进行参数优选,建立了基于GA优化的GA-Gaussian-SVR高速公路短时交通量预测模型,将本路段前几个时段交通量、天气因素和出行日期作为影响因素输入,结合实例进行了仿真预测.结果表明该方法可有效应用于高速公路短时交通量预测.

Aiming at real-time,volatility and nonlinearity of the freeway short-term traffic flow prediction,support vector regression was used to forecast the short-term traffic flow of freeway.The ε-insensitive loss function of support vector regression was replaced with Gaussian loss function to deal with the random error of normal distribution effectively in the original sequence.To seek the optimal unknown parameters of Gaussian-SVR,Genetic algorithm was applied to choose the parameters of Gaussian-SVR,a new freeway traffic volume forecasting model named by GA-Gaussian-SVR was proposed.In actual forecasting,several pre-period traffic flow of this road section,weather and date were taken into account.Contrastive analysis was made with instance data,analysis results show that this model can be effectively applied to prediction of the freeway traffic volume.