Fast computing method for two-stage robust network constrained Unit Commitment problem

Unit Commitment (UC) problem plays a key role in power systems' operations. Due to the strong modeling capacity of robust optimization to capture various complicated uncertainties in power systems, including those in renewable generation, load realization, demand response, and contingencies, two-stage robust UC has gained intensive attentions from both academic scholars and industry practitioners. However, such tri-level optimization problem is very challenging to solve, considering large-scale real power grids. In this research, we study new computational methods and strategies to address this challenge, including incorporating strong formulations for basic UC model, deriving new valid inequalities considering network constraints, and designing and implementing a decomposition procedure. On large-scale test instances, our solution approach leads to significantly better computational performance, compared to existing formulations or methods.