Forecast of traffic speeds with neural network LSTM encoder-decoder
DOI:
https://doi.org/10.14295/transportes.v30i3.2660Keywords:
Congestion, Traffic forecasting, Neural networksAbstract
This article proposes a speed prediction model for a highway segment in the city of Porto Alegre, which has daily traffic jams due to bottlenecks. We used traffic data and environmental variables, such as rainfall intensity, accidents and atypical events to make the forecasts. Then we proposed a neural network model with an encoder-decoder architecture and long short-term memory (LSTM) layers, which has the characteristic of establishing long-term relationships between the input variables, being relevant for applications in the Transportation area. As additional contributions, we evaluated the quality of forecasts for different prediction horizons and traffic regimes. We compared cumulative distribution functions (CDFs) generated using field and forecast data using a survival analysis method similar to the breakdown probability calculation. These CDFs represent the probability of a sudden speed drop due to the transition from the free-flow to the congested regime. The methodology presented a satisfactory performance based on both criteria, making good predictions even in critical traffic situations.
Downloads
References
do Amaral, M.B. (2020) Previsão de velocidades de tráfego com rede neural LSTM. Bachelor thesis. Departmento de Engenharia Civil: Universidade Federal do Rio Grande do Sul. Porto Alegre, Brazil.
Andrade, G.R. and Setti, J.R. (2014) Speed–Flow Relationship and Capacity for Expressways in Brazil, in Innovative Applications of the Highway Capacity Manual 2010. DOI: http://onlinepubs.trb.org/onlinepubs/circulars/ec190.pdf.
Arras, L. et al. (2019) Explaining and Interpreting LSTMs, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 11700 LNCS(2019), pp. 211–238. DOI: 10.1007/978-3-030-28954-6_11.
Brilon, W., Geistefeldt, J. and Regler, M. (2005) Reliability of Freeway Traffic Flow: A stochastic Concept of Capacity, Proceedings of the 16th International Symposium on Transportation and Traffic Theory, (July), pp. 125–144.
Caleffi, F. et al. (2016) Influência das condições climáticas e de acidentes na caracterização do comportamento do tráfego em rodovias, Transportes, 24(4), p. 57. DOI: 10.14295/transportes.v24i4.1104.
Caleffi, F. (2018) Proposição de um método de harmonização da velocidade baseado em modelo de previsão de conflitos veiculares. PhD Thesis. Departamento de Engenharia de Produção e Transportes: Universidade Federal do Rio Grande do Sul. Porto Alegre, Brazil.
Chollet, F. (2018) Deep Learning with Python. 1st edn. Shelten Island, NY: Manning Publications.
Elefteriadou, L. et al. (2011) Proactive ramp management under the threat of freeway-flow breakdown, Procedia - Social and Behavioral Sciences, 16, pp. 4–14. DOI: 10.1016/j.sbspro.2011.04.424.
Elefteriadou, L. et al. (2014) Enhancing ramp metering algorithms with the use of probability of breakdown models, Journal of Transportation Engineering, 140(4), pp. 1–9. DOI: 10.1061/(ASCE)TE.1943-5436.0000653.
Fortunato, M., Blundell, C. and Vinyals, O. (2017) Bayesian Recurrent Neural Networks, pp. 1–14. DOI: 10.48550/arXiv.1704.02798.
Fu, R., Zhang, Z. and Li, L. (2017) Using LSTM and GRU neural network methods for traffic flow prediction, Proceedings - 2016 31st Youth Academic Annual Conference of Chinese Association of Automation, YAC 2016, pp. 324–328. DOI: 10.1109/YAC.2016.7804912.
Gal, Y. and Ghahramani, Z. (2016) A theoretically grounded application of dropout in recurrent neural networks, Advances in Neural Information Processing Systems, pp. 1027–1035. DOI: 10.48550/arXiv.1512.05287.
Galvan, Y.T., Zechin, D. and Cybis, H.B.B. (2019) Utilização do método Kaplan-Meier na estimativa das distribuições de velocidade desejada, Anais do 33º Congresso de Pesquisa e Ensino em Transportes - 2019. Rio de Janeiro: Associação Nacional de Pesquisa e Ensino em Transportes.
Géron, A. (2019) Hands-On Machine Learning with Scikit-Learn, Keras & TensorFlow. 2nd edn. Sebastopol: O’Reilly Media
Gu, Y. et al. (2019) Short-term prediction of lane-level traffic speeds: A fusion deep learning model, Transportation Research Part C: Emerging Technologies, 106(July), pp. 1–16. DOI: 10.1016/j.trc.2019.07.003.
Han, Y. and Ahn, S. (2018) Stochastic modeling of breakdown at freeway merge bottleneck and traffic control method using connected automated vehicle, Transportation Research Part B: Methodological, 107, pp. 146–166. DOI: 10.1016/j.trb.2017.11.007.
Hochreiter, S. and Schmidhuber, J. (1997) Long Short-Term Memory, Neural Computation, 9(8), pp. 1735–1780. DOI: 10.1162/neco.1997.9.8.1735.
Iung, B. (2013) Cœur et grossesse, EMC - Traité de médecine AKOS, 8(2), pp. 1–4. DOI: 10.1016/s1634-6939(13)59289-1.
Kendall, A. and Gal, Y. (2017) What uncertainties do we need in Bayesian deep learning for computer vision?, Advances in Neural Information Processing Systems, 2017-December(Nips), pp. 5575–5585. DOI: 10.48550/arXiv.1703.04977.
Laptev, N. et al. (2017) Time-series Extreme Event Forecasting with Neural Networks at Uber, International Conference on Machine Learning - Time Series Workshop, pp. 1–5. Available at: <http://www.cs.columbia.edu/~lierranli/publications/TSW2017_paper.pdf>. Accessed at: 12/08/2022.
Li, L. et al. (2018) Hyperband: A novel bandit-based approach to hyperparameter optimization, Journal of Machine Learning Research, 18, pp. 1–52. DOI: 10.48550/arXiv.1603.06560.
Li, Y., Wang, N. and Carroll, R.J. (2010) Generalized functional linear models with semiparametric single-index interactions, Journal of the American Statistical Association, 105(490), pp. 621–633. DOI: 10.1198/jasa.2010.tm09313.
Ma, X. et al. (2015) Long short-term memory neural network for traffic speed prediction using remote microwave sensor data, Transportation Research Part C: Emerging Technologies, 54, pp. 187–197. DOI: 10.1016/j.trc.2015.03.014.
Ma, Y., Zhang, Z. and Ihler, A. (2020) Multi-Lane Short-Term Traffic Forecasting with Convolutional LSTM Network, IEEE Access, 8, pp. 34629–34643. DOI: 10.1109/ACCESS.2020.2974575.
Neil, D., Pfeiffer, M. and Liu, S.C. (2016) Phased LSTM: Accelerating recurrent network training for long or event-based sequences, Advances in Neural Information Processing Systems, (Nips), pp. 3889–3897. DOI: 10.48550/arXiv.1610.09513.
Pujol, B.S., Asplund, M. and Serra, J.C. (2019) Machine learning for early detection of traffic congestion using public transport traffic data. Bachelor Thesis. Department of Computer and Information Science Machine: Linköping University. Linköping, Sweden.
Vlahogianni, E.I., Karlaftis, M.G. and Golias, J.C. (2014) Short-term traffic forecasting: Where we are and where we’re going, Transportation Research Part C: Emerging Technologies, 43, pp. 3–19. DOI: 10.1016/j.trc.2014.01.005.
Zechin, D., Caleffi, F. and Cybis, H.B.B. (2020) Influence of Rain on Highway Breakdown Probability, Transportation Research Record, 2674(8), pp. 687–695. DOI: 10.1177/0361198120919754.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2022 Douglas Zechin, Matheus Basso do Amaral, Helena Beatriz Bettella Cybis
This work is licensed under a Creative Commons Attribution 4.0 International License.
Authors who submit papers for publication by TRANSPORTES agree to the following terms:
- Authors retain copyright and grant TRANSPORTES the right of first publication with the work simultaneously licensed under a Creative Commons Attribution License that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
- Authors may enter into separate, additional contractual arrangements for the non-exclusive distribution of this journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in TRANSPORTES.
- Authors are allowed and encouraged to post their work online (e.g., in institutional repositories or on their website) after publication of the article. Authors are encouraged to use links to TRANSPORTES (e.g., DOIs or direct links) when posting the article online, as TRANSPORTES is freely available to all readers.
- Authors have secured all necessary clearances and written permissions to published the work and grant copyright under the terms of this agreement. Furthermore, the authors assume full responsibility for any copyright infringements related to the article, exonerating ANPET and TRANSPORTES of any responsibility regarding copyright infringement.
- Authors assume full responsibility for the contents of the article submitted for review, including all necessary clearances for divulgation of data and results, exonerating ANPET and TRANSPORTES of any responsibility regarding to this aspect.
