Sustainable Mobility and Energy Ecosystem for Transportation Electrification

Abstract

The transition to electric mobility is not a set of independent changes but a coupled process in which transportation demand, charging infrastructure, and energy use evolve together. When these elements are studied separately, the mechanisms linking them remain unclear, limiting the ability to design effective policies and infrastructure strategies. This thesis develops an integrated framework that follows this transition from demand formation to system interaction and, ultimately, to infrastructure decision-making. The analysis begins with transportation energy demand, where a hybrid forecasting approach is constructed by combining machine learning with a structured representation of energy sources. This formulation allows the model to capture nonlinear demand patterns while preserving consistency with the underlying energy system, so that predicted trends reflect both behavioral change and shifts in energy demand and 〖CO〗_2 emissions. The demand representation is then refined through a behavioral layer, where charging and travel patterns are examined using a redesigned deep clustering model. By replacing the conventional encoder with a variational structure and embedding feature-weighted learning with an optimization-based determination of cluster structure, the method identifies stable user groups and shows how seasonal conditions alter charging needs across these groups. These demands and behavioral patterns are subsequently embedded within a system-level structure that treats electric vehicle adoption, charging infrastructure, and transport energy demand as mutually dependent processes. A triangular dynamic framework is developed to identify how changes propagate across these components without imposing a predefined causal order. By introducing stage-conditioned identification, explicit differentiation between directional propagation and feedback, and stability-constrained inference, the analysis reveals that system behavior shifts across regimes rather than following a fixed pathway, with these shifts shaped by spatial interaction, policy conditions, and external shocks. This system understanding is then translated into infrastructure planning through a multi-objective stochastic-robust optimization model. The model determines charger deployment strategies that remain feasible under uncertain demand while optimizing cost, user time, and lifecycle emissions within realistic technological and spatial constraints. The resulting framework establishes a coherent progression from demand to behavior, from behavior to system dynamics, and from system dynamics to infrastructure decisions, providing a consistent basis for planning electrified transportation systems.