Citation

Ma R, Chen S, Zhang HM. J. Intell. Transp. Syst. 2021; 25(4): 423-438.

Copyright

(Copyright © 2021, Informa - Taylor and Francis Group)

DOI

10.1080/15472450.2021.1927728

PMID

unavailable

Abstract

This paper investigates time-series correlations between macroscopic travel speed and parking garage occupancy in downtown area, using the real-time parking occupancy data via SFPark.org and travel speed data from HERE Maps for San Francisco downtown areas as a data-driven case study. This study significantly expands recent work on instantaneous correlations by incorporating variables as time series. The equivalency between the nonlinear regression with logistic curves and the single-node single hidden layer neural network is established. By testing time delay neural network models, this study investigates the time delay effects between macroscopic travel speed and parking garage occupancy. The performance of single-layer multi-nodes nonlinear autoregressive with exogenous inputs neural network is evaluated, which suggests such types of time series neural networks can effectively forecast macroscopic travel speed by using travel speed and parking occupancy information with various forecasting delay tabs.

Language: en

Keywords

Parking; prediction; time delay neural networks; time series; traffic speed