Home // VEHICULAR 2021, The Tenth International Conference on Advances in Vehicular Systems, Technologies and Applications // View article

Hybrid Neural Network Modeling for Multiple Intersections along Signalized Arterials - Current Situation and Some New Results

Authors:
Wan Li
Chieh Wang
Yunli Shao
Hong Wang
Guohui Zhang
Tianwei Ma
Jon Ringler
Danielle Chou

Keywords: Signalized intersections; modeling; neural networks; performance analysis; signalized arterials simulation.

Abstract:
Traffic flow along signalized arterials is a dynamic, nonlinear, and stochastic system in which the relationship between the signal timing plan and traffic delays is too complicated to be modeled using first principles approaches. With advances in sensing technologies, various data sets are available, allowing effective data-driven modeling to be conducted for further controller design. In this keynote paper, a Hybrid Neural Network (HNN) is proposed to model the multiple intersections along a signalized arterial in Honolulu, in which both modeling structure and the relevant training algorithms have been developed. HNN modeling using real data has shown a set of promising results, with dynamic model performance assessed using model error Probability Density Function (PDF). A simple HNN model can easily be used as a starting point for an artificial intelligence–based closed-loop control design that controls the signal timing to reduce the traffic delay.

Pages: 10 to 20

Copyright: Copyright (c) IARIA, 2021

Publication date: July 18, 2021

Published in: conference

ISSN: 2327-2058

ISBN: 978-1-61208-879-2

Location: Nice, France

Dates: from July 18, 2021 to July 22, 2021