A branch-and-check approach for a wind turbine maintenance scheduling problem
| Autor: | Louis-Martin Rousseau, Aurélien Froger, Michel Gendreau, Jorge E. Mendoza, Eric Pinson |
|---|---|
| Přispěvatelé: | Laboratoire Angevin de Recherche en Ingénierie des Systèmes (LARIS), Université d'Angers (UA), ERL CNRS 6305 Ordonnancement et Conduite, E3i, Université de Tours (UT)-Université de Tours (UT), Université de Tours-Université de Tours |
| Rok vydání: | 2017 |
| Předmět: |
Schedule
Mathematical optimization 021103 operations research General Computer Science Job shop scheduling Computer science Testbed 0211 other engineering and technologies [INFO.INFO-RO]Computer Science [cs]/Operations Research [cs.RO] 02 engineering and technology Management Science and Operations Research Time limit 7. Clean energy Task (project management) Electricity generation Modeling and Simulation 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing Electric power industry Integer programming |
| Zdroj: | Computers and Operations Research Computers and Operations Research, Elsevier, 2017, 88, pp.117-136. ⟨10.1016/j.cor.2017.07.001⟩ |
| ISSN: | 0305-0548 |
| Popis: | International audience; In this paper we deal with a maintenance scheduling problem arising in the onshore wind power industry. We consider a short-term horizon and a multi-skilled workforce. The goal is to schedule maintenance operations to maximize electricity production while taking into account forecast wind-speed values, multiple task execution modes, and daily restrictions on the routes of the technicians. We first introduce two integer linear programming formulations of the problem. Then, building on one of our models, we propose a branch-and-check (B&C) approach that uses both generic Benders cuts and cuts specially crafted for our problem. We report experiments on a 160-instance testbed. For 80% of the instances, our exact approach finds an optimal solution in a reasonable computational time. The remaining instances reach the three-hour time limit, and our B&C gives solutions with average gaps of 1.7% with respect to the upper bounds. The results suggest that our method significantly outperforms commercial solvers running our integer linear programming models. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full Text from ScienceDirect