摘要
针对传统方法由于未能在交通流预测前对危险弯坡组合路段线形数据进行恢复,导致预测误差大、预测时间长等问题,提出极端天气下行车危险弯坡组合路段交通流预测方法。方法通过路段中心轴点的提取,完成路段线形数据的恢复;再通过对极端天气类型的描述,获取极端天气衡量指标并完成天气数据的筛选;将决策树与随机森林法相结合,建立交通流预测模型,最后将恢复的路段线形数据以及筛选的天气数据放入模型中,实现极端天气下危险弯坡组合路段交通流预测。实验结果表明,运用上述方法预测交通流时,预测误差低、预测时间短。
Aiming at the problems of large prediction error and long prediction time caused by the failure of traditional methods to recover the linear data of dangerous curve and slope combination sections before traffic flow prediction,a traffic flow prediction method for dangerous curve and slope combination sections under extreme weather is proposed.This method restored linear data by extracting central axis points of the road section.Through the description for extreme weather types,the measurement indexes of extreme weather were obtained,and the data screening was completed.Moreover,decision tree was combined with random forest method to build a model of predicting traffic flow.Finally,the linear data of the restored road section and the weather data were added to the model.Thus,the traffic flow prediction of dangerous curved combination road section in extreme weather was achieved.Experimental results prove that the proposed method has low prediction error and short prediction time.
作者
张瑾
字丰军
ZHANG Jin;ZI Feng-jun(Kunming University of Science and Technology,Faculty of Transportation Engineering,Kunming Yunnan 650000,China)
出处
《计算机仿真》
北大核心
2022年第11期179-183,共5页
Computer Simulation
关键词
极端天气
危险弯坡
组合路段
随机森林
交通流预测
Extreme weather
Dangerous curved slope
Combination road sections
Random forest
Traffic flow prediction