Virtual Power Plant Energy Coordination: A Global Solution to Three-Level Scheduling

In robust energy scheduling for virtual power plants (VPPs), uncertainty sets (USs) have been used to capture the uncertainty in renewable energy source (RES) outputs. However, for VPP energy scheduling models that employ methods such as the convex hull US (CHUS), which involves a large number of integer variables, traditional column-and-constraint generation (C&CG) algorithms often converge to local optima, posing significant challenges for practical scheduling. This necessitates the development of algorithms capable of escaping local optima. To balance security and economic efficiency, this study adopts a confidence level-based CHUS, which improves scheduling economics by setting a specified confidence level. Subsequently, a vertex-based column-and-constraint generation algorithm is designed to avoid local optima, enabling the confidence level-based CHUS to solve the robust VPP energy scheduling model effectively. Finally, based on an actual VPP system in South China, the effectiveness and superiority of the proposed US method and algorithm are validated.