SHORT TERM TRAFFIC AND POLLUTION PREDICTION ON THE UK MOTORWAY NETWORK



SHORT TERM TRAFFIC AND POLLUTION PREDICTION ON THE UK MOTORWAY NETWORK

Authors

Carl Goves, Transport Systems Catapult, Fabio Galatioto, Transport Systems Catapult, Abraham Narh, Transport Systems Catapult

Description

This paper presents the results of integrating innovative prediction engines to estimate traffic conditions and pollution exceedances in the short term (up to 2 hours) on a section of the M60 motorway within the UK.

Abstract

The ability to predict combined traffic and pollution levels over a short-term period has been rarely explored in the past. The potential to improve traffic and air quality management by allowing a proactive approach, rather than reactive, can lead to more effective decisions and strategies being employed. This could include enabling automatic traffic management such as variable speed limits in advance of expected traffic congestion or undesirable pollution levels.

This paper provides an advancement on early findings on two separate research streams namely traffic and pollution short-term prediction. It explores how the combination of multiple techniques, including the application of artificial intelligence, may lead to a more accurate prediction of both traffic conditions and pollution levels. The case study has been chosen on part of the UK motorway network, specifically the M60, where traffic and air quality impacts are important.

Preliminary results show that the new approach can predict the correct traffic congestion state 90% of the time for 15 minute predictions. This leads to predictions of pollution exceedances above the legal limits being improved by 25% compared to previous research.

The deployment of short term traffic predictions using live data feeds is also explored which enables the possibility of producing ‘traffic casts’ of future traffic conditions. The presence of corrupted data in real time data feeds is identified as a significant barrier to accurate short term predictions. However, preliminary results show predictions remain reliable as long as no more than 15% of the input data is corrupted.

The applied research conducted as part of this study forms part of a rolling programme being conducted by the Transport Systems Catapult, the UK’s innovation centre for Intelligent Mobility. In addition, this paper forms a natural progression to the research awarded the Neil Mansfield Award at the European Transport Conference in 2015 to the principal author.

Publisher

Association for European Transport