Resilient pricing-driven coordination of coupled power and transportation networks: A learning-augmented mixed-integer reformulation approach

Large-scale electric vehicle (EV) integration brings new solutions to enhance resilience in coupled power and transportation networks (CPTNs). However, the inherent complexity of co-optimizing power distribution and transportation networks poses significant challenges, and their interdependencies under resilient dispatch remain unclear. To address these problems, this work develops a resilient pricing-driven bilevel coordination approach to jointly optimize power distribution network (PDN) restoration and transportation network (TN) price-aware dynamic traffic and energy assignment (DTEA). At the upper level, the PDN operator determines coordinated restoration and resilient V2G price decisions. At the lower level, a novel price-aware DTEA model is introduced that simultaneously optimizes EV traffic and energy flows to fully exploit spatio-temporal V2G flexibility. A data-driven constraint learning-based method is further proposed to reformulate nonlinear DTEA into a tractable accurate mixed-integer program (MIP). Unlike black-box surrogates, the proposed method learns nonlinear constraints via neural networks (NNs) and explicitly embeds the learned NN structures into the optimization model. A value-function-based algorithm, supplemented by tailored linearization techniques, is developed to solve the resulting bilevel MIP. Comprehensive case studies demonstrate that the proposed approach improves CPTN resilience through pricing-driven coordination, achieves superior restoration performance by leveraging V2G flexibility, and attains favorable scalability and computational efficiency.