Optimality-feasibility-aware multistage unit commitment considering nonanticipative realization of uncertainty

Abstract

The intrinsic uncertainty in source-load power and line outage poses huge obstacles to unit commitment (UC) optimization in power grids. To satisfy the practical requests in power grids, this paper exploits a multistage robust UC method to promise both optimality and feasibility in nonanticipative scheduling. Firstly, a multistage robust UC model is established accounting for the sequential realization of uncertainty. This model makes decisions with the goals of economy in the normal condition and power balance in the emergent condition. Secondly, a customized multi-cut decomposition algorithm is offered to address the intractable multistage robust optimization problem. The original multistage problem is decoupled into the form of one master and several slave problems (MP-SPs), where the robust dual dynamic programming (RDDP) solves the resulting multistage max–min SPs, and the column-and-constraint generation (C&CG) algorithm successively constructs multiple cutting planes to ensure the optimality and convergence of the alternative optimizations between the MP and SPs. Computational tests on real-size power grids validate the practicability and superiority of the proposed multistage robust UC method, which is of great significance in guiding system operations.