基于最小二乘支持向量机( Least squares support vector machine ,LS-SVM)算法建立符合我国道路交通流特征的车辆跟驰模型,并用该模型模拟单车道道路上车辆的跟驰行为。采用NGSIM提供的数据对LS-SVM模型进行仿真验证,将测试结果与传统的Gipps模型进行对比。结果表明:与Gipps模型相比,LS-SVM模型对应的各项误差指标精度均有明显改善,能够挖掘变量之间的潜在关系,弥补传统车辆跟驰模型的不足。
Car-Following Behavior Model Based on Least Squares Support Vector Machine
The least squares support vector machine ( LS-SVM) algorithm describes and fits the car following behaviors on roads with Chinese traffic flow characteristics .The study uses the LS-SVM model to simulate the car-following behaviors on single-lane road.It uses NGSIM data to verify LS-SVM model and also tests the model against traditional Gipps model .The results show that the accuracy of error indicators of the LS-SVM model is improved more significantly than the Gipps model .The LS-SVM model is able to display the potential relationships between variables and compensate the shortage of traditional car-following model .
car-followingmachine learningleast squares support vector machineregression forecast