Abstract
This paper discusses a two-stage fully distributed approach for the unit commitment (UC) problem, which solves the UC of each individual unit. This approach, based on the augmented Lagrangian relaxation method (ALR) and alternating direction method of multipliers (ADMM), consists of two stages: the solution of the single-unit optimization problem and the distributed update of the multiplier. In the first stage, using the proximal term to effectively improve the oscillation phenomenon and the convergence speed of the improved distributed augmented Lagrangian relaxation method (ID-ALR). In the second stage of multiplier updating, using ADMM to solve dual subgradient problem in fully distributed way for obtaining subgradients. In addition, the multiplier is updated with the improved multiplier update strategy. Finally, the subgradients are also used to reduce the constraint violation degree of the UC and enhance the soft constraint ability of the ID-ALR. The simulation results show that the ID-ALR can not only obtain high-quality solutions in reasonable times but also obtain good speedup in a parallel environment. The proposed ID-ALR enables each unit to keep its information secret and achieves a fully distributed solution for UC.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call