摘要
文中提出了一种基于路段关联度的城市交通流量Apriori-LSTM(Apriori-long short term memory network)预测模型.处理卡口检测器数据,统计交通量并提取车辆轨迹,采用Apriori算法分析预测时段内目标路段与关联路段的时空相关性,计算关联路段支持度;求解预测时段内关联路段到目标路段的流入量,构建LSTM预测的输入矩阵、并采用LSTM预测路段短时流量.采用实例进行验证,对迭代次数、隐藏层神经元个数和步长进行参数灵敏度分析,并与单一的LSTM预测结果进行比较.结果表明:Apriori-LSTM的平均绝对误差降至3.8%,平均绝对百分误差和平均均方误差均有显著降低,均等系数有所提高,模型稳定性更好,达到了更好预测效果.
An Apriori-LSTM(Apriori-Long Short Term Memory Network)forecasting model of urban traffic flow based on link correlation degree was proposed.By processing the data of the bayonet detector,the traffic volume is counted and the vehicle trajectory was extracted.Apriori algorithm was used to analyze the temporal and spatial correlation between the target road section and the associated road section in the prediction period,and the support degree of the associated road section was calculated.The inflow from the associated link to the target link in the forecast period was solved,the input matrix of LSTM prediction was constructed,and the short-term flow of the link was predicted by LSTM.Through the example verification,the parameter sensitivity of iteration times,number of hidden layer neurons and step size were analyzed,and the results were compared with the single LSTM prediction results.The results show that the average absolute error of Apriori-LSTM is reduced to 3.8%,the average absolute percentage error and mean square error are significantly reduced,and the equalization coefficient is improved.The stability of the model is better,and the prediction effect is better.
作者
陈玲娟
杨任泉
CHEN Lingjuan;YANG Renquan(School of Automobile and Traffic Engineering,Wuhan University of Science and Technology,Wuhan 430065,China)
出处
《武汉理工大学学报（交通科学与工程版）》
2023年第3期402-407,共6页
Journal of Wuhan University of Technology（Transportation Science & Engineering）
基金
教育部人文社会科学研究项目(19YJCZH007)。
