计算机科学

计算机科学 ›› 2024, Vol. 51 ›› Issue (11A): 231000180-7.doi: 10.11896/jsjkx.231000180

• 大数据&数据科学 • 上一篇    下一篇

FCTNet:基于双域深度学习的公交车到站时间预测方法

张铭泽1,2, 李轶1, 吴文渊1, 石明全1, 王正江3   

  1. 1 中国科学院重庆绿色智能技术研究院自动推理与认知重庆市重点实验室 重庆 400714
    2 中国科学院大学重庆学院 重庆 400714
    3 重庆市公共交通控股集团凤筑科技有限公司 重庆 401120
  • 出版日期:2024-11-16 发布日期:2024-11-13
  • 通讯作者: 李轶(zm_liyi@163.com)
  • 作者简介:(zhangmingze21@mails.ucas.ac.cn)
  • 基金资助:
    重庆市院士牵头科技创新引导专项(cstc2020yszx-jcyjX0005,cstc2021yszx-jcyjX0005,cstc2022YSZX-JCX0011CSTB)

FCTNet:Bus Arrival Time Prediction Method Based on Dual Domain Deep Learning

ZHANG Mingze1,2, LI Yi1, WU Wenyuan1, SHI Mingquan1, WANG Zhengjiang3   

  1. 1 Chongqing Key Laboratory of Automated Reasoning and Cognition,Chongqing Institute of Green and Intelligent Technology,Chinese Academy of Science,Chongqing 400714，China
    2 Chongqing School,University of Chinese Academy of Sciences,Chongqing 400714,China
    3 Fengzhu Technology Co.,Chongqing Public Traffic Holdings Group,Chongqing 401120,China
  • Online:2024-11-16 Published:2024-11-13
  • About author:ZHANG Mingze,born in 1997,master.His main research interests include time series prediction and deep learning.
    LI Yi,born in 1980,Ph.D,associatc professor.His main research interests include program verification,symbolic computation and intelligent transportation.
  • Supported by:
    Chongqing Academician-led Science and Technology Innovation Guidance Special Project(cstc2020yszx-jcyjX0005,cstc2021yszx-jcyjX0005,cstc2022YSZX-JCX0011CSTB).

PDF (PC)

170

摘要： 公交车到站时间预测是智能公交系统的重要组成部分,可以给乘客提供精确的到站时间,还可以帮助调度员进行更合理的调度安排。为此,提出一种基于卷积、注意力机制和FFT的对时域和频域进行双域深度学习的公交车到站时间预测算法FCTNet(FFT-Conv-Transformer),该算法融合了傅里叶变换、卷积神经网络和注意力机制,其可以对公交车单站和多站的到站时间进行预测。其中利用傅里叶变换和卷积神经网络在频域上学习输入数据的时空特征,同时保留时域信号,利用注意力机制学习输入序列的全局依赖关系,预测最终结果。在重庆市465,506和262这3条公交线路到站时间预测实验中,FCTNet网络模型的平均绝对百分比误差和平均绝对误差优于实验对比算法,在最繁忙的465线路中FCTNet网络模型的平均相对误差相对已有最好模型降低了2.34%,平均绝对误差降低了4.59s。

关键词: 到站预测, 注意力机制, 时域频域转换, 卷积神经网络, 深度神经网络

Abstract: Bus arrival time prediction is an important part of the smart bus system.It can provide passengers with accurate arrival time and help dispatchers make more reasonable scheduling arrangements.A bus arrival time prediction algorithm FCTNet(FFT-Conv-Transformer)based on convolution,attention mechanism and FFT is proposed for dual-domain deep learning of time domain and frequency domain.The algorithm integrates Fourier transform,convolutional neural network and attention mechanism to predict bus arrival times at single stops and multiple stops.Among them,Fourier transform and convolutional neural network are used to learn the spatiotemporal characteristics of the input data in the frequency domain while retaining the time domain signal.The attention mechanism is used to learn the global dependence of the input sequence and predict the final result.In the arrival time prediction experiment of three bus lines 465,506 and 262 in Chongqing,the average absolute percentage error and average absolute error of the FCTNet network model are better than the experimental comparison algorithm.In the busiest bus No.465,the average relative error of the FCTNet network model is better than that of the experimental comparison algorithm.Compared with the existing best model,it is reduced by 2.34%,and the average absolute error is reduced by 4.59s.

Key words: Arrival forecast, Attention mechanism, Time-frequency transformation, Convolution neural network, Deep neural network

中图分类号: 

  • TP183
[1]XU H,YING J.Bus arrival time prediction with real-time and historic data [J].Cluster Computing,2017,20(4):3099-3106.
[2]GONG J,LIU M,ZHANG S.Hybrid dynamic prediction modelof bus arrival time based on weighted of historical and real-time GPS data[C]//2013 25th Chinese Control and Decision Confe-rence(CCDC).IEEE,2013:972-976.
[3]ČELAN M,LEP M.Bus-arrival time prediction using bus net-work data model and time periods [J].Future Generation Computer Systems,2020,110:364-371.
[4]BO C,PENG Z R.Bus travel time prediction based on dynamicmodel[J].Computer Engineering and Application,2016,52(3):103-107.
[5]WANG B,WANG W,YANG M,et al.An approach to bus tra-vel time prediction based on the adaptive fading Kalman filter algorithm [C]//Multimodal Transportation Systems－Conve-nient,Safe,Cost-Effective,Efficient(CICTP 2012).2012:1652-1661.
[6]ABIDIN A F,KOLBERG M.Towards Improved Vehicle ArrivalTime Prediction in Public Transportation:Integrating SUMO and Kalman Filter Models [C]//2015 17th UKSim-AMSS International Conference on Modelling and Simulation(UKSim).2015:147-152.
[7]YIN T,ZHONG G,ZHANG J,et al.A prediction model of busarrival time at stops with multi-routes [J].Transportation Research Procedia,2017,25:4623-4636.
[8]REDDY K K,KUMAR B A,VANAJAKSHI L.Bus travel time prediction under high variability conditions [J].Current Science,2016:700-711.
[9]COMI A,ZHUK M,KOVALYSHYN V,et al.Investigating bus travel time and predictive models:a time series-based approach [J].Transportation Research Procedia,2020,45:692-699.
[10]PETERSEN N C,RODRIGUES F,PEREIRA F C.Multi-output bus travel time prediction with convolutional LSTM neural network [J].Expert Systems with Applications,2019,120:426-435.
[11]MAZLOUMI E,ROSE G,CURRIE G,et al.Prediction intervals to account for uncertainties in neural network predictions:Me-thodology and application in bus travel time prediction [J].Engineering Applications of Artificial Intelligence,2011,24(3):534-542.
[12]BAI C,PENG Z R,LU Q C,et al.Dynamic Bus Travel TimePrediction Models on Road with Multiple Bus Routes [J].Comput Intell Neurosci,2015:432389.
[13]LIU Z Z.Research on Bus Arrival time Prediction Method based on GRU and Kalman Filter combination model [J].China-arab States Science and Technology Forum(Chinese and English),2022,(9):100-104.
[14]LAI Y X,ZHANG L,YANG F,et al.Bus Arrival Time Prediction Algorithm Based on Spatio-temporal Correlation Attribute Model [J].Journal of Software,2020,31(3):648-662.
[15]ZHANG X H,LIU H J,SHI J Q,et al.Urban Bus Travel Time Prediction Based on Long and Short Term Memory Artificial Neural Network Based on Spatial-temporal Feature Vector [J].Journal of Computer Applications,2021,41(3):875-880.
[16]VASWANI A,SHAZEER N,PARMAR N,et al.Attention IsAll You Need[J].arXiv:1706.03762,2017.
[17]YANG H M,PAN Z S,BAI W.Review of time series prediction methods [J].Computer Science,2019,46(1):21-28.
[18]LECUN Y,BOTTOU L,BENGIO Y,et al.Gradient-basedlearning applied to document recognition [J].Proceedings of the IEEE,1998,86(11):2278-2324.
[19]ZHOU H,ZHANG S,PENG J,et al.Informer:Beyond efficient transformer for long sequence time-series forecasting[C]//Proceedings of the AAAI Conference on Artificial Intelligence,2021.
[20]WU H,XU J,WANG J,et al.Autoformer:Decomposition transformers with auto-correlation for long-term series forecasting [J].Advances in Neural Information Processing Systems,2021,34:22419-22430.
[21]ZHOU T,MA Z,WEN Q,et al.Fedformer:Frequency enhanced decomposed transformer for long-term series forecasting[C]//Proceedings of the International Conference on Machine Lear-ning.PMLR,2022.
[22]LAI G,CHANG W C,YANG Y,et al.Modeling long-and short-term temporal patterns with deep neural networks[C]//Proceedings of the The 41st International ACM SIGIR Conference on Research & Development in Information Retrieval.2018.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市渝北区洪湖西路18号    邮编：401121  
电话：023-63500828（编辑部） 023-67039612（会议/专题） 023-67039623（财务/发票）
E-mail：jsjkx12@163.com
本系统由北京玛格泰克科技发展有限公司设计开发