摘要
北京市交通运行指数是定量反映北京市路网拥堵状态的重要指标,交通运行状态预测是构建智慧交通系统的重要研究内容.在模式序列匹配算法(PSF)基础上优化预测序列权重,针对交通运行指数的时序相关性,增加了基于反距离权重的时间衰减因子,提高了临近序列在交通运行模式匹配计算中的重要性.实验结果表明:与传统时间序列模型(ARIMA)、深度学习模型(LSTM)和标准模式序列匹配算法进行对比分析,改进的模式序列匹配预测算法有着较高的预测精度,且具有较强的自适应性.
Beijing Traffic Performance Index(TPI)is an important index to quantitatively reflect the congestion state of road network in Beijing,and traffic operation state prediction is an important research content of constructing intelligent traffic system.In this paper,based on the Pattern Sequence Forecasting(PSF)algorithm,the prediction sequence weight is optimized.Aiming at the temporal correlation of the TPI,the time attenuation factor based on the Inverse Distance Weighted(IDW)is added to improve the importance of the adjacent sequence in the traffic operation pattern matching calculation.The experimental results show that:compared with the traditional time series model(ARIMA),the deep learning model(LSTM)and the general PSF algorithm,the improved PSF algorithm has higher prediction accuracy and stronger adaptability.
作者
郭小刚
张健钦
卢剑
陆浩
李卓航
GUO Xiaogang;ZHANG Jianqin;LU Jian;LU Hao;LI Zhuohang(School of Geomatics and Urban Information,Beijing University of Civil Engineering and Architecture,Beijing 100044)
出处
《北京建筑大学学报》
2019年第4期20-28,共9页
Journal of Beijing University of Civil Engineering and Architecture
基金
国家自然科学基金项目(41771413,41701473)
关键词
交通运行指数
时序预测
模式序列匹配
反距离权重
traffic performance index(TPI)
time series prediction
pattern sequence forecasting(PSF)algorithm
inverse distance weighted(IDW)
