Robust Scheduling Strategy for Low-Carbon EV Charging with Photovoltaic Uncertainty

This paper proposes a dynamic low-carbon charging strategy to address the spatiotemporal heterogeneity of grid carbon intensity caused by high-penetration renewable integration and large-scale EV adoption. First, we develop an improved carbon intensity model based on thermal-renewable coordination, incorporating Beta-distributed PV output interval prediction to quantify the impacts of power mix, load characteristics and transmission constraints on carbon dynamics. Second, we design a two-layer “day-ahead robust planning and real-time rolling correction” optimization framework where the upper layer generates baseline schedules resilient to carbon intensity prediction errors through box-type uncertainty sets, while the lower layer dynamically adjusts charging power based on real-time observations guided by a tiered carbon pricing mechanism for load shifting. Case studies demonstrate the proposed strategy reduces load forecasting errors by $15\%$ during midday PV fluctuations, shrinks peak-valley differences to 0.8 MW, increases PV utilization to $82\%$ via a dual-mode energy storage strategy, and lowers total system costs by $12.3\%$, providing a robust and cost-effective solution for source-loadstorage coordination in next-generation power systems.