A quick center-point algorithm for unit commitment with carbon emission trading. (September 2020)

• Record Type: Journal Article
• Authors: Yang, Linfeng; Li, Wei; Chen, Guo; Fang, Beihua; Tang, Chunming
• Is Part Of: International journal of electrical power & energy systems. Volume 120(2020)
• Publication Date: 2020-09
• Subjects: Unit commitment -- Center-point algorithm -- Outer approximation -- Perspective-cut -- Integer ellipsoid center; Electrical engineering -- Periodicals; Electric power systems -- Periodicals; Électrotechnique -- Périodiques; Réseaux électriques (Énergie) -- Périodiques; Electric power systems; Electrical engineering; Periodicals; 621.3
• DOI: 10.1016/j.ijepes.2020.105996
• Languages: English
• ISSNs: 0142-0615

Abstract

Highlights: We propose a novel global optimization algorithm to quickly solve the unit commitment problem with carbon emission trading. We innovatively introduce the integer ellipsoid center to improve the performance of the Center-Point algorithm. The Center-Point algorithm can find high-quality solutions significantly faster compared to CPLEX. Abstract: A novel global optimization algorithm, named center-point (CP) algorithm, is proposed to solve the unit commitment problem with carbon emission trading (UC-CET) in this paper. By solving a sequence of linear continuous-relaxed subproblems and performing line search procedures, the CP algorithm constructs a sequence of perspective-cuts to generate a tight linear approximation of UC-CET. Then the algorithm iteratively finds integer ellipsoid center of the current linear approximation as the trial solutions, and makes the trial solutions close to the optimal solution by searching the neighborhood of these solutions and adding new linear constraints. We build two types of UC-CET models, one of which considers power balance constraints and the other considers DC power flow constraints. Then, based on these models, we compared the CP algorithm with state-of-the-art solver CPLEX. The simulation results show that the proposed algorithm can find high-quality solutions significantly faster than CPLEX and it is suitable to solve large-scale UC-CET problem.