Probabilistic traffic breakdown forecasting through Bayesian approximation using variational LSTMs

Abstract

This paper proposes a framework for short-term traffic breakdown probability calculation using a Variational LSTM neural network model. Considering that traffic breakdown is a stochastic event, this forecast framework was devised to produce distributions as outputs, which cannot be achieved using standard deterministic recurrent neural networks. Therefore, the model counts on the robustness of neural networks but also includes the stochastic characteristics of highway capacity. The framework consists of three main steps: (i) build and train a probabilistic speed forecasting neural network, (ii) forecast speed distributions with the trained model using Monte Carlo (MC) dropout, and therefore perform Bayesian approximation, and (iii) establish a speed threshold that characterizes breakdown occurrence and calculate breakdown probabilities based on the speed distributions. The proposed framework produced an efficient control over traffic breakdown occurrence, can deal with many independent variables or features, and can be combined with traffic management strategies.

KEYWORDS:

• Traffic breakdown
• traffic forecasting
• neural networks
• Bayesian statistics
• machine learning

Acknowledgements

This research is supported by grants from CAPES (Coordination of Superior Level Staff Improvement), Brazil.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by CAPES: [Grant Number ].

Notes on contributors

Douglas Zechin

Douglas Zechin Doctoral candidate and Master in Transport Systems at the Federal University of Rio Grande do Sul (UFRGS) and graduated in Civil Engineering at the same university with an emphasis on Transport Systems and Civil Structures. He did an exchange program at the Technical University of Munich (Germany) focused on Transport Systems and participated in the HyperloopTT pre-feasibility study at Serra Gaúcha, Brazil. He is experienced in traffic simulation, active traffic management, and machine learning applications in both fields.

Helena Beatriz Bettella Cybis

Helena Beatriz Bettella Cybis Graduated in Civil Engineering at the Federal University of Rio Grande do Sul (1980), Master in Transport - University of Leeds (1989), Ph.D. in Transport - University of Leeds (1993), post-doctorate at the University of Berkeley (2010). She is a full professor at the Department of Production and Transport Engineering at the Federal University of Rio Grande do Sul. She was president of the National Association for Research and Education in Transport from 2017 to 2020, having been Scientific Director and President of the Scientific Committee of the Association's annual congresses from 2007 to 2008 and 2011 to 2014 and Vice-president of the areas of Traffic Engineering and Safety of the Pan American Congress of Traffic Engineering, Transport and Logistics from 2007 to 2014. She has experience in transport engineering and transport planning, acting on the following topics: traffic engineering, traffic allocation models, traffic models and traffic simulation, and the study of pedestrian behaviour.