Traffic‐Aware Comanagement in Urban Mobility Grids: Quantile Forecasts and Cyber Attack‐Defensive Scheduling

This paper presents an integrated, cyber‐aware energy management model for interconnected electric vehicle (EV) and metro systems. This model specifically addresses two key components: (i) the static scheduling of EV connections at parking hubs and depots, and (ii) dynamic control of traction energy in response to fluctuating traffic rates and routing policies. A multimodal GAN‐LSTM‐CNN forecasting model is utilized to generate high‐resolution load profiles for EV hubs and metro traction power. This forecaster incorporates various factors, such as traffic intensity and historical charging data. It employs a conditional generative adversarial network (GAN) to produce unusual traffic and charging patterns, a long short‐term memory (LSTM) network to capture long‐term temporal effects, and convolutional neural network (CNN) layers to detect local patterns and feature interactions. This approach results in sharper and well‐calibrated profiles that are resilient to distribution shifts and minor forecast manipulations. Given that intelligent coordination increases potential vulnerabilities, we develop a malware threat model that attempts to manipulate forecasts and actuation commands. In response, we implement a defense‐in‐depth strategy that includes authenticated control, a moving‐target defense mechanism for charger grouping and keys, and a physics‐informed congruence monitor that rejects any commands or forecasts violating power‐flow and energy‐budget constraints. In a metropolitan case study involving multiple traction substations and EV parking hubs, our proposed forecaster reduces prediction error by up to 76% compared with recurrent quantum neural networks (RQNs) and support vector regression (SVR) across typical and stress scenarios, such as incidents and event days. This improvement results in significantly lower peaks in feeder demand, reduced energy costs, and fewer minutes of overload or violations. Our findings demonstrate that a collaboratively designed approach to forecasting, optimization, and cyber resilience can greatly enhance the efficiency and reliability of electric urban mobility.