Multi-modal Mesoscopic Transportation System Modeling and Management with Mobility Data

2019-10-09T20:57:45Z (GMT) by Xidong Pi
A multi-modal transportation system is a transportation system with multiple travel modes, which usually includes solo-driving, car-pooling, public transit, freight transportation, ride-sourcing, park-and-ride, etc. With emerging technologies, multimodal transportation becomes highly ubiquitous and diversified nowadays. From the system perspective, the co-existence of all these travel modes lead to diversified and complex systems that enables more instruments to alleviate traffic congestion and improve users’ quality of life, provided that the comprehensive systems are managed and operated in a proper way. To properly manage a multi-modal transportation system, we need a holistic
modeling framework of multi-modal transportation networks to support the decision making, including tasks like transportation demand management, what-if analysis,
congestion alleviation, estimation of system costs, etc. However, due to the complexity of the multi-modal transportation system, there are several challenges for
building such a model. First, traffic in such systems usually contains different types of vehicles. How to model the behavior and estimate the travel cost of these vehicles
in the heterogeneous traffic flow remains unsolved. Secondly, it is also challenging to integrate different travel modes in a single multi-modal transportation system.
Their linkage and interdependency is complicated and stochastic due to the nature of human’s behaviors.
In this dissertation, we focus on the problem of multi-modal transportation system modeling and management leveraging mobility data. The final goal is to comprehensively
model the entire multi-modal transportation system, to better understand the interaction and linkage among each component of transportation systems, and to ultimately facilitate optimal decision making on urban transportation system planning and operation. Specifically, we propose a simulation-based dynamic traffic assignment model
for cars and trucks to simulate the traffic flow in the system. We build a public transit operation performance analytics framework using multiple types of mobility
data. We also propose a model to optimize the curbside parking of ridesourcing vehicles. Finally, a holistic multi-modal dynamic user equilibrium model is proposed
to integrate all travel modes and model the travelers’ behaviors in a general multimodal transportation system. Numerical experiments based on different multi-modal
transportation networks are conducted for each of the work and show our models and methods have excellent performance and applicability. In this dissertation work, we mainly focus on the day-to-day equilibrium state of
the multi-modal transportation systems. Other scenarios including abnormal, hazardous and disaster situations are not considered. Moreover, the calibration of the
multi-modal dynamic traffic assignment model is also not the main focus of this dissertation, which is studied and introduced in other papers of the author and collaborators.