基于概率电压灵敏度分析与霍尔定理的主动配电网中移动式储能系统路由与调度

Abstract Mobile energy storage systems (MESSs) possess significant temporal and spatial flexibility, making them ideal for ancillary services in active distribution networks (ADNs). However, conventional MESS scheduling methods rely heavily on accurate load and traffic forecasts, while deep learning-based approaches can be computationally expensive and insufficiently adaptive to dynamic system conditions. To address these challenges, we propose a two-stage scheduling framework that integrates sensitivity analysis, graph theory, and dynamic optimization techniques, thereby enhancing adaptability and computational efficiency. In the first stage, a destination pre-generation model leverages probabilistic voltage sensitivity to accommodate load forecast uncertainties and pinpoint critical ADN nodes that are most likely to require ancillary support. In the second stage, an innovative destination screening algorithm based on Hall's theorem refines the candidate nodes, coupled with a dynamic rolling optimization scheme that continuously updates MESS routes and charging/discharging strategies in real-time. Numerical simulations demonstrate that, compared to existing methods, our proposed two-stage framework improves scheduling accuracy by 5.56 %, boosts the mission finish rate by 35.27 %, and extends the average hourly duration of ancillary services by roughly 20 min. These results underscore the framework's effectiveness and adaptability, offering a robust solution for reliable ADN operations.