Spatio-Temporal Clustering for Non-Recurrent Traffic Congestion Detection on Urban Road Networks
Non-Recurrent Congestion events (NRCs) frustrate commuters, companies and traffic operators because they cause unexpected delays. Most existing studies consider NRCs to be an outcome of incidents on motorways. The differences between motorways and urban road networks, and the fact that incidents are not the only cause of NRCs, limit the usefulness of existing automatic incident detection methods for identifying NRCs on an urban road network. This thesis contributes to the literature by developing an NRC detection methodology to support the accurate detection of NRCs on large urban road networks. To achieve this, substantially high Link Journey Time estimates (LJTs) on adjacent links that occur at the same time are clustered. Substantially high LJTs are defined in two different ways: (i) those LJTs that are greater than a threshold, (ii) those LJTs that belong to a statistically significant Space-Time Region (STR). These two different ways of defining the term ‘substantially high LJT’ lead to different NRC detection methods. To evaluate these methods, two novel criteria are proposed. The first criterion, high-confidence episodes, assesses to what extent substantially high LJTs that last for a minimum duration are detected. The second criterion, the Localisation Index, assesses to what extent detected NRCs could be related to incidents. The proposed NRC detection methodology is tested for London’s urban road network, which consists of 424 links. Different levels of travel demand are analysed in order to establish a complete understanding of the developed methodology. Optimum parameter settings of the two proposed NRC detection methods are determined by sensitivity analysis. Related to the first method, LJTs that are at least 40% higher than their expected values are found to maintain the best balance between the proposed evaluation criteria for detecting NRCs. Related to the second method, it is found that constructing STRs by considering temporal adjacencies rather than spatial adjacencies improves the performance of the method. These findings are applied in real life situations to demonstrate the advantages and limitations of the proposed NRC detection methods. Traffic operation centres could readily start using the proposed NRC detection methodology. In this way, traffic operators could be able to quantify the impact of incidents and develop effective NRC reduction strategies.
- Anbaroglu B., Cheng T.; Where and when does the traffic congestion begin and end? A spatio-temporal clustering approach to detect congestion, Proceedings of ISPRS Workshop on Spatio-Temporal Data Mining and Analysis, July 18-19, 2011, London.
- Cheng T., Anbaroglu B.; Spatio-Temporal Clustering of Road Network Data; Artificial Intelligence and Computational Intelligence (AICI), 2010.
- Cheng T., Anbaroglu B.; Defining Dynamic Spatio-Temporal Neighbourhood of Network Data, 14th Int. Symposium on Spatial Data Handling (ISGIS), 2010.
- Cheng T., Anbaroglu B.; Methods on Defining Spatio-Temporal Neighbourhoods; GeoComputation, 2009.
- Cheng T., Anbaroglu B.; Spatio-Temporal Outlier Detection in Environmental Data, Spatial and Temporal Reasoning for Ambient Intelligence Systems Workshop, 2009.
Page last modified on 17 dec 13 14:28