Urban public transport systems are playing an increasingly important role in sustainable urban development. However, with cities’ growing dependency on public transport (e.g., metro, bus), even slight disruptions to the system by adverse weather, accidents, special and extreme events can lead to widespread and cascading chaos and significant productivity loss. Many large cities around the world have been suffering from unplanned disruptions of large urban transportation systems due to weather extremes, metro track intrusions, and rolling stock failures. On the other hand, premeditated large-scale man-made events—such as major sporting events, maintenance closure—may also cause significant impact and inconvenience to travelers. In light of the negative effects of disruptions and events, it becomes urgent for cities and transport agencies to understand the impact of disruptions on passengers and develop effective and efficient response measures.
This research project addresses the aforementioned disruption management and response problems for multi-modal transportation systems to enhance the resilience of integrated transportation systems and optimize recovery-oriented response operations during and after disruptions. Specifically, the project aims to (1) understand travelers’ behavior in integrated multi-modal transit systems under disruptions due to extreme and special events based on emerging data (e.g., smart card and social media), and (2) design efficient and effective recovery-oriented response operations to mitigate the negative impact of disruptions with the aid of machine learning, operations research, data-driven optimization, and other relevant advanced techniques.
The proposed research project consists of four modules: (1) robust and flexible learning of travel demand estimation under disruptions: We will devise a Bayesian deep neural network-based framework for spatio-temporal travel demand prediction and flow anomaly detection. We will also study the event-driven joint modeling of traffic flow and passenger behavior for risk warning to incorporate the external event information into the model for more effective early prediction for risk management. (2) integrated metro and bus network design for resilience enhancement: We will design different disruption management schemes for the integrated metro and bus system (e.g., running bus bridging/paratransit services, adjusting metro timetable) for post-disruption recovery, and propose integrated network design to enhance the system’s resilience against disruptions from a pre-event perspective. (3) pricing and dispatching of e-hailing service system in extreme/special events: We will study problems regarding pricing and incentive mechanisms to shed light on optimal management of the e-hailing and on-demand systems and thus to improve the service quality and efficiency in extreme and special events. (4) post-event emergency management and preventive resilience enhancement for public charging infrastructure: Recognizing the vulnerability of the infrastructure to the cyberattack, we attempt to improve the resistance and resilience of the charging-facility network by conducting investigations along three threads, i.e., impact prediction, preventive measure, and post-event emergency management.
To validate the theories and methods outlined in the foregoing paragraph, real-world case studies based on the cities of Montreal and Shanghai will be conducted to gain managerial insights for the operation and management of urban public transport systems. This research will provide timely decision support to public transport agencies/operators as for operating and planning for future smart and resilient cities.