A multi-objective distributionally robust chance-constrained model for power grid resilience enhancement with limited offensive information

Abstract

Frequent deliberate attacks have greatly affected the security of power grids. Defense resources, such as mobile emergency generators (MEGs), have been deployed to enhance the resilience of the grid. However, the uncertainty of offensive resource information poses a serious challenge to developing defense strategies. This paper introduces a multi-objective distributionally robust chance-constrained (MODRCC) method for planning the pre-storage capacity of MEGs, aiming to address the uncertainty of the number of attack resources and enhance grid resilience. A combinatorial optimization framework for planning MEG storage and generation capacity is developed, and the operational state of the attacked grid and the MEG scheduling scheme are modeled. Subsequently, the Distributed Robust Chance Constraints (DRCC) approach is developed to address the uncertainty of offensive resources. A deterministic reformulation over Wasserstein balls is used to convert DRCC method into mixed-integer conic programs that can be solved directly by commercial solvers. Furthermore, a non-dominated sorting genetic algorithm-II (NSGA-II) solution is designed to continuously update the MEG storage scheme based on the expected load shedding obtained by the DRCC. Finally, case studies are conducted on the IEEE 24-bus and 118-bus systems to verify the effectiveness of the proposed method.