Search

Intelligent Transport Systems (ITS) is a field that has developed rapidly over the last two decades, driven by the growing need for better transport network management strategies and by continuing improvements in computing power. However, a number of ITS applications, such as Advanced Traveller Information Systems (ATIS), Dynamic Route Guidance (DRG) and Urban Traffic Control (UTC) need to be proactive rather than reactive, and consequently require the prediction of traffic state variables into the short-term future. Similarly, individual travellers can use this predictive information to plan their mobility more efficiently. This PhD thesis develops models that are able to accurately predict short-term traffic variables such as link travel time and traffic flow on urban arterial roads under both normal and abnormal traffic conditions. This research first reviews the state of the art in data prediction applications in engineering domains especially traffic engineering and presents existing statistical and machine learning methods and their applications in relation to short-term traffic prediction. This review establishes that most existing work has focused on the apparent superiority of one individual statistical or machine learning method over another. Little attention has been paid, however, to the issues surrounding the overall structure of prediction models, in particular in relation to data smoothing and error feedback. In developing a short-term traffic prediction model, therefore, a 3-stage framework including a data smoothing step and an error feedback mechanism is proposed. This proposed framework is applied in conjunction with five different machine learning methods to develop a range of short-term traffic prediction methods. The proposed prediction framework is then tested under different traffic conditions using traffic data generated from a traffic simulation model of a corridor in Southampton. The prediction results show that the proposed 3-stage prediction framework can improve the accuracy of traffic prediction, regardless of the machine learning method used under both normal and abnormal traffic conditions. After demonstrating the effectiveness of predicting traffic variables using simulated data, the proposed methodology is then applied to real-world traffic data collected from different sites in London and Maidstone. These results also show that the framework can improve the accuracy of prediction regardless of the machine learning tool used. The prediction accuracy comparison shows that the proposed 3-stage prediction framework can improve the prediction accuracy for either travel time or traffic flow data under both normal and abnormal traffic conditions. In addition, the results indicate that the kNN based prediction method, when applied through the proposed framework, outperforms other selected machine learning methods under abnormal traffic conditions on urban roads. The findings suggest that, in order to arrive at a robust and accurate prediction model, attention should be paid to combining data smoothing, model structure and error feedback elements. Open Access

Autonomous Vehicles (AVs) are bringing revolutionary opportunities and challenges to urban transport systems. They can reduce congestion, improve operational efficiency and liberate drivers from driving. Though AVs might bring attractive potential benefits, most benefits are evaluated at high AV penetration rates or an all-AV scenario. In practice, limited by price barriers, adoption rates and vehicle-renewal periods, AVs may not replace Human-Driven Vehicles (HDVs) to achieve a high penetration rate in a short time. It can be expected that the road network will operate with a mix of AVs and HDVs in the near to medium future. Therefore, there is a strong motivation to analyse the performance of road networks under mixed traffic conditions. The overall aim of this PhD research is to analyse mixed traffic flows of AVs and HDVs to help traffic managers and Local Authorities (LAs) improve the performance of urban traffic systems by right-of-way reallocation and dynamic traffic management. To achieve this aim, this PhD research is divided into four parts. Firstly, the impact of heterogeneity between AVs and HDVs on road capacity is investigated. A theoretical model is proposed to calculate the maximum capacity of heterogeneous traffic flow. Based on the theoretical model, it is shown that road capacity increases convexly with AV penetration rates. This finding provides a theoretical basis to support the hypothesis that right-of-way reallocation can increase road capacity under the mixed traffic flow. To cross-validate the above finding, different right-of-way reallocation strategies are evaluated on a two-lane road with SUMO simulation. Compared with a do-nothing scenario, the road capacity can be increased by approximately 11% with a proper RoW reallocation strategy at low or medium AV penetration rates. Secondly, whether CAVs can be used as mobile traffic controllers by adjusting their speed on a certain link is investigated. It is found that in some circumstances, system efficiency can be improved by CAVs adjusting their speed on a certain link to nudge the network towards the system optimum. According to a numerical analysis on the Braess network, total travel time can be reduced by 9.7% when CAVs actively slow down on a link. To take more realistic circumstances into account, a SUMO simulation case study is conducted, where HDVs only have partial knowledge about travel costs. The results of the simulation demonstrate that when CAVs are acting as mobile traffic controllers by actively reducing speed on a certain link, total travel time can be reduced by approximately 6.8% compared with the do-nothing scenario. Thirdly, whether travel efficiency can be improved with only a part of the vehicular flow cooperatively changing their routing under mixed conditions is investigated. It has been found that it is possible to use CAVs to influence HDVs’ day-to-day routing and push the network towards the system optimal distribution dynamically on a large network with multiple OD pairs. Taking non-linear cost-flow relationship and signal timing into account, an Optimal Routing and Signal Timing (ORST) control strategy is proposed for CAVs and tested in simulation. Compared with initial user equilibrium, total travel time can be reduced by approximately 7% when a portion of CAVs cooperatively charge their routing with the ORST control strategy at the 75% CAV penetration rate. This opens up possibilities, besides road pricing, to improve system efficiency by controlling routing and signal timing strategy for CAVs. Fourthly, whether additional travel efficiency can be achieved by jointly optimising routing and signal timing with information from CAVs is further investigated. Specifically, the impact of information levels on routing and signal timing efficiency has been investigated quantitatively. The results demonstrate that different levels of information will lead the road traffic system to reach different equilibrium points. Then the proposed ORST control strategy is compared with existing routing and signal timing strategies. The results present that ORST can reduce approximately 10% of the total travel time compared to user equilibrium. In addition, the proposed model has also been tested on a revised Nguyen-Dupuis network. At 25% CAV penetration rates, the proposed model can successfully reduce approximately 23% of total travel time. In summary, the mixed flow of AVs and HDVs is investigated in this PhD research. To increase the efficiency of urban traffic systems, novel strategies have been proposed and tested with numerical analysis and simulation, which provides inspirations and quantitative evidence for traffic managers and LAs to manage the mixed traffic flow efficiently. Open Access