摘要
从时间和空间2个维度分析了交通流速度的自相关和互相关特性,提出了交通流速度的时空二维预测模型,可对未来交通流进行多步预测。该模型包含曲线拟合加权预测模型和递推时空二维自回归预测模型,动态模型参数可由递推最小二乘法实时识别,静态模型参数通过离线优化得到。将预测的交通流速度应用于车辆宏观运动规划方法中,可将车辆的燃油经济性进一步提高,在上下班高峰路段,油耗可进一步降低近10%。
The autocorrelation and cross-correlation characteristics of traffic flow speed were analyzed from two dimensions of time and space;a two-dimensional spatiotemporal prediction model of traffic flow speed was proposed,which can be used for multi-step prediction of future traffic flow.This model includes periodic weighted average model and recursive two-dimensional spatiotemporal autoregressive prediction model,the dynamic model parameters were identified by recursive least square method in real-time,and the static model parameters were optimized off line.The fuel economy of vehicles can be further improved by applying the predicted traffic flow speed to the vehicle macroscopic motion planning method,and the fuel consumption can be further reduced by nearly 10%in rush hour sections.
作者
苗成生
刘海鸥
Miao Chengsheng;Liu Haiou(Automotive Research&Development Center,Guangzhou Automotive Group Co.Ltd,Guangzhou 511434,China;Beijing Institute of Technology,Beijing 100081,China)
出处
《湖北汽车工业学院学报》
2021年第3期42-49,53,共9页
Journal of Hubei University Of Automotive Technology
基金
国家重点研发计划项目(2017YFB0102600)。
关键词
交通流
时空二维预测模型
递推时空二维自回归预测方法
递推最小二乘法
traffic flow
two-dimensional spatiotemporal prediction model
recursive two-dimensional spatiotemporal autoregressive prediction method
recursive least square method
