A Data-Driven Simulation-Based Method for Transportation Network Design

This research develops a data-driven simulation-based method, which integrates high-resolution simulators and machine learning techniques, to model and solve the NDP. Three major tasks are tackled: a) developing a computable traffic flow model for representing traffic dynamics in interrupted multimodal environments, b) developing a generic data-driven simulation-based approach for NDP with a specific focus on network topology, by establishing a simulation-based bi-level model and a solution algorithm based on Bayesian optimization, c) incorporating practical and political constraints into the optimization of the network capacity analysis. The proposed framework can serve as a decision-making tool for transport planners and traffic engineers.