In 2012, the NZTA appointed Sinclair Knight Merz (SKM) to carry out research into Travel Time Predictability. This research intends to clarify how historical base line data combined with near real-time data including environmental conditions, incidents and traffic flow can contribute to the calculation of reliable and timely-delivered travel time predictions. Reliable journey time is a key parameter in travellers’ route choice and has important applications in transport planning and modelling. For transport users, it affects their choice of mode, journey route and also their activity patterns. For transport planners and policy makers, journey time estimates are used to provide key indicators for performance monitoring, congestion management, travel demand modelling and forecasting, traffic simulation, air quality analysis, evaluation of travel demand and traffic operations strategies. Real time travel time information is becoming increasingly important for a variety of transport applications - including Advanced Traveller Information Systems (ATIS), Advanced Traffic Management Systems (ATMS), Route Guidance Systems (RGS), etc., which all form part of the collective Intelligent Transportation System (ITS). As Intelligent Transportation Systems (ITS) are deployed more widely throughout the world, managers of transport systems have increasing access to large amounts of historical and ‘‘real-time’’ status data. Traffic flows, speeds and densities on transport network are being continuously measured by different monitoring systems such as loop detectors, Automatic Number Plate Recognition systems (ANPR), Closed Circuit Television (CCTV) monitoring and, more recently, probe vehicles and mobile phone data. The collected information can be used to guide the use of dynamic traffic management measures such as variable message signs, information provision by radio and internet, etc. in order to reduce congestion and improve network efficiency. The availability of real-time traffic information, developments in Information Technology and the need for predicting short-term traffic conditions have raised the question: Can we predict travel time? It has been recognized by researchers and practitioners alike that the benefits of ITS capabilities cannot be fully realised without an ability to anticipate traffic conditions in the short-term (i.e. less than one hour into the future). Predictive modelling capability can theoretically provide the ability to forecast the performance of a transport network in the short term and may also allow the impact of planned and unplanned events and incidents on the network to be assessed in near real-time. The key objectives of the research were to: Understand the current state of research locally and internationally on real-time prediction methodology — specifically identify from current research the most suitable methodology for predicting journey times from real-time data sources for strategic highway networks. Understand what data sources currently exist and identify the gaps in the current information sources which can lead to accurate journey time predictions. Develop summary statistics of the observed data to identify the impact of key parameters on journey times in ‘normal’ traffic conditions and also during ‘abnormal’ conditions resulting from planned and unplanned events and incidents Develop an appropriate forecasting model to predict travel times with a reasonable degree of confidence and develop a testing methodology for verifying the results Summarise the findings in a comprehensive research report which also provides a programme for delivery of a Travel Time Predictive Capability for the strategic highway network. The research needs to deliver a modelling framework that can forecast journey times based on near realtime traffic information and historic data sources. The modelling approach should attempt to take into account all explanatory variables which will affect the reliability of forecast results. The model will also need to incorporate Spatio-Temporal relationships which have a direct impact on traffic on the highway sections where journey time is being forecast. One of the most import aspects of the research is to understand existing data sources and gaps in traffic information which can affect the reliability of modelling results. In addition, the research will also focus on the impact of planned and unplanned disruptions in the highway network capacity. (Author/publisher)
Abstract