A Fast Solution Method for Large-Scale Unit Commitment Based on Virtual Power Boundary

Abstract

This paper investigates the fast solution difficulty for large-scale unit commitment (UC) problems. The UC problem is a typical puzzle in the scheduling of power systems. Lagrangian relaxation (LR) and mixed-integer linear programming (MILP) are the two most widely adopted methods. However, in the LR framework, constructing feasible solutions and updating Lagrangian multipliers are two major challenges, and the efficiency of MILP methods is restricted by the scale of UC problems. To this end, this paper proposes a new fast solution method for large-scale UC based on virtual power boundary. Specifically, first, an initial trial solution is obtained based on the established state transition diagram. Second, a feasibility test model based on virtual power boundary is established to check the feasibility of trial UC solutions effectively. Third, the updating strategies for UC solutions are proposed based on the feasible gap solved by the feasibility test model. The procedure is implemented until a good feasible UC solution is obtained. Numerical results implemented on IEEE 24-bus, IEEE 31-bus, IEEE 118-bus, and Polish 2383-bus systems show the efficiency of the proposed method.