计算机工程 ›› 2019, Vol. 45 ›› Issue (6): 1-5.doi: 10.19678/j.issn.1000-3428.0052120

所属专题: 智能交通专题

• 智能交通专题 • 上一篇    下一篇

基于维度加权的残差LSTM短期交通流量预测

李月龙1,a3,唐德华1,a2,3,姜桂圆2,肖志涛1,b3,耿磊1,b3,张芳1,b3,吴骏3   

  1. 1. 天津工业大学 a.计算机科学与技术学院; b.电子与信息工程学院,天津 300387
    2. 南洋理工大学 计算机科学与工程学院,新加坡 639798
    3. 天津市光电检测技术与系统重点实验室,天津 300387
  • 收稿日期:2018-07-16 修回日期:2018-08-31 出版日期:2019-06-15 发布日期:2019-06-17
  • 作者简介:李月龙(1982—),男,副教授、博士,主研方向为智能交通、机器学习、模式识别|唐德华,硕士研究生;姜桂圆,研究员、博士;肖志涛,教授、博士|肖志涛,教授、博士;耿 磊、张 芳、吴 骏,副教授、博士
  • 基金项目:
    国家自然科学基金(61771340,61302127,61601325);天津市自然科学基金(18JCYBJC15300)

Short Term Traffic Flow Forecasting Based on Dimension Weighted Residual LSTM

Yuelong LI1,a3,Dehua TANG1,a2,3,Guiyuan JIANG2,Zhitao XIAO1,b3,Lei GENG1,b3,Fang ZHANG1,b3,Jun WU3   

  1. 1. a.School of Computer Science and Technology;b.School of Electronics and Information Engineering,Tianjin Polytechnic University,Tianjin 300387,China
    2. School of Computer Science and Engineering,Nanyang Technological University,Singapore 639798,Singapore
    3. Tianjin Key Laboratory of Optoelectronic Detection Technology and Systems,Tianjin 300387,China
  • Received:2018-07-16 Revised:2018-08-31 Online:2019-06-15 Published:2019-06-17

摘要:

基于神经网络的交通流量预测由于嵌入了部分手工设计的特征,使得提取的网络特征功能单一,存在适应性及鲁棒性差、数据局部特征刻画不准确等问题。为此,提出基于残差长短期记忆网络(LSTM)的交通流量预测方法,利用集成学习思想将空间分布的数据端到端训练到残差LSTM网络中,同时在每个LSTM单元后引入维度加权单元,显式建模特征维度之间的相互依赖关系。实验结果表明,该方法能实现短期交通流量数据的自适应建模分析。

关键词: 智能交通, 短期交通流量预测, 残差连接, 长短期记忆网络, 维度加权

Abstract:

Due to the fact that the current neural network-based traffic flow forecasting method embeds part of the manually designed features,the feature extracted by the network has single function,bad adaptability,poor robustness and inaccurate characterization of the local features.Therefore,this paper proposes a traffic flow forecasting method based on residual Long Short Term Memory(LSTM).The model uses the idea of ensemble learning to train the spatially distributed data into a residual LSTM network.In addition,dimension weighted units are introduced after each LSTM unit to present the interdependencies between different dimensions of modeling features.Experimental results show that the method can realize adaptive modeling analysis of short term traffic flow data.

Key words: intelligent transportation, short term traffic flow forecasting, residual connection, Long Short Term Memory(LSTM) network, dimension weighted