A real-time online decision support system for intermodal passenger travel.

PDFTech Reporthttp://www.ncitec.msstate.edu/wp-content/uploads/2013-13FR.pdfNCITEC Project No. 2013-13Decision support systemsReal time informationIntermodal transportationUnited StatesMississippi State University. National Center for Intermodal Transportation for Economic CompetitivenessHu, MengqiWu, Yao-JanChen, YangYang, ShuMississippi State University. National Center for Intermodal Transportation for Economic CompetitivenessUS Transportation CollectionThe transportation system in the United States is disjointed and inefficient as a result of thedifferent transportation modes in use and their respective industries which have developedindependently. In addition, public transportation is not well used in passenger trips compared toother developed countries. For example, public transportation accounts for 20% of passenger tripsin large U.S. cities compared to 50% in Europe. Also, development of the passenger intermodaltransportation system has lagged behind development of the freight transportation system.To improve utilization of intermodal transit and efficiency in the U.S, we developed anintelligent decision support system for passenger travel decisions using real-time general transitfeed specifications (GTFS) data. In our system, an automatic data collection strategy was createdto collect GTFS and flight data across different platforms, and an \ue2\u20ac\u153all-in-one\ue2\u20ac? database wasdesigned to store the data. The database was used to: 1) construct intermodal transit networks usinga \ue2\u20ac\u153node-link\ue2\u20ac? scheme, and 2) estimate travel time and travel time reliability for links and transitroutes. Using this real-time data, a data-driven travel decision model was developed to determinethe best route based on passenger preferences. Several chance constraints were added in thedecision model to guarantee the reliability of the travel route under uncertainties. Additionally, auser-friendly interface was developed in Python to allow travelers to plan their trips, and ageographic information system (GIS), Google Earth, was employed to allow users to visualize theoptimized route options.The proposed system was validated using real-time GTFS data collected in Tucson, AZ, andBoston, MA. This validation demonstrated that the system can determine optimal travel routes forpassengers. In addition, three sets of sensitivity analysis experiments were developed to investigatethree model considerations: 1) the effect of chance constraints on path choice, 2) the effect ofconfidence levels on path choice, and 3) the difference between weekend and weekday travelplanning. The results suggested that the optimal anticipated travel time increases with anincreasing on-time arrival confidence level, and walking is preferred by passengers instead oftransferring buses during peak hours. As an example, approximately 30% additional time servesas a reference for allocating travel buffer time to ensure a higher on-time arrival confidence levelfor transit trips to the Tucson International Airport