Inactive Constraint Identification for Unit Commitment by Neural Network

Unit commitment is crucial for optimizing scheduling in power systems and electricity markets. However, because it is typically modeled as a large-scale mixed-integer linear programming problem, its computational efficiency is limited, restricting its further application in the power industry. One significant factor that makes the unit commitment problem challenging is its dense constraints, particularly the numerous power flow constraints due to the transmission limits of the power grid. This paper addresses the challenge by integrating datadriven methods and using neural networks to identify inactive transmission constraints, simplifying the unit commitment model. The proposed method is validated on the RTS-GMLC system, and the results demonstrate that it can effectively identify inactive constraints and improve the computational efficiency of the unit commitment problem.