Constrained non-linear multi-objective optimisation of preventive maintenance scheduling for offshore wind farms

• Published in: Mechanical systems and signal processing 2018-05, Vol.104, p.347-369
• Main Authors: Zhong, Shuya, Pantelous, Athanasios A., Beer, Michael, Zhou, Jian
• Format: Article
• Language: English
• Subjects: Classifying; Cost parameters; Genetic algorithms; Maintenance; Maintenance management; Mathematical programming; Multi-objective Programming; Multiple objective analysis; Offshore energy sources; Offshore wind farms; Optimization; Pareto optimization; Power supplies; Preventive maintenance; Reliability; Schedules; Scheduling; Sorting algorithms; System reliability; Turbines; Wind farms; Wind power; Wind power generation
• ISSN: 0888-3270; 1096-1216
• DOI: 10.1016/j.ymssp.2017.10.035

•A non-linear multi-objective programming model is proposed for preventive maintenance of offshore wind farms.•Maximization of system reliability and minimisation of maintenance related to cost are considered simultaneously.•The optimisation is solved with a nondominated sorting genetic algorithm II. Offshore wind farm is an emerging source of renewable energy, which has been shown to have tremendous potential in recent years. In this blooming area, a key challenge is that the preventive maintenance of offshore turbines should be scheduled reasonably to satisfy the power supply without failure. In this direction, two significant goals should be considered simultaneously as a trade-off. One is to maximise the system reliability and the other is to minimise the maintenance related cost. Thus, a non-linear multi-objective programming model is proposed including two newly defined objectives with thirteen families of constraints suitable for the preventive maintenance of offshore wind farms. In order to solve our model effectively, the nondominated sorting genetic algorithm II, especially for the multi-objective optimisation is utilised and Pareto-optimal solutions of schedules can be obtained to offer adequate support to decision-makers. Finally, an example is given to illustrate the performances of the devised model and algorithm, and explore the relationships of the two targets with the help of a contrast model.