Prediction of Traffic Speed and Volume Considering Malfunctioning Detectors by Using Deep Learning Model
To keep abreast of the latest traffic information is important for drivers and the public. This research aims to develop a deep learning model for predicting traffic speed and volume within the coming one hour when some detectors malfunction. The Deep Learning model is also applicable for imputing missing data in offline applications. Clustering methods are used to reduce the sample size and the cost of training the deep learning model, in addition to improving the prediction accuracy of the entire traffic network.
A ready-to-use software prototype will be developed to provide more accurate traffic state prediction, offering drivers a better navigation service with a more accurate prediction of traffic congestion and travel time. It can also assist the transport agencies in more effective traffic management, traffic flow control and signal control.
The project has received support from the Smart Traffic Fund.
(Smart Traffic Fund is funding initiated by the Transport Department to support local organisations or enterprises for conducting research and application of innovation and technology with the objectives of enhancing commuting convenience, enhancing the efficiency of the road network or road space, and improving driving safety)
