Efficient Affine Adjustable Robust Unit Commitment With Generalized Ellipsoidal Uncertainty Sets

Authors:

Abstract
Robust optimization is widely used for unit commitment problems with uncertain variables, where the definition of uncertainty sets and equivalence transformation of robust models critically affect solution conservativeness and computational efficiency. This paper proposes a novel approach to construct minimal generalized ellipsoidal uncertainty sets by selecting representative sample points using Mahalanobis distance, thereby eliminating redundant regions. To address the computational challenge posed by second-order cone constraints in adjustable robust models, a piecewise-affine policies is introduced to map generalized ellipsoidal sets to convex hulls, simplifying the feasible region and reducing solution complexity. Case studies on the IEEE 118-bus systems demonstrate that the proposed method reduces uncertainty set volume by 37.45%, respectively, and decreases computation time by 99.69% compared with direct equivalence transformation.