To ensure that wave power reaches its full potential it is important to optimise all aspects of the technology. The optimisation process requires us to consider computationally heavy simulations and several objective functions, so one should carefully choose which optimisation algorithm is most suitable. This study has reviewed three different multi-objective optimisation algorithms: NSGA-II, MO-CMA-ES and MOPSO. The algorithms will optimise a wave park in respect to its generated power, power fluctuation, cost and park area. Multi-objective optimisation results in a so-called Pareto front of many optimal solutions, and this study has investigated how to choose one preferred solution from the Pareto front to best satisfy the user's requirements. The results show that NSGA-II and MOPSO are fast algorithms that can reliably converge towards non-dominated solutions, although NSGA-II may miss essential parts of the solution space and MOPSO is reliant on uncertain parameters. MO-CMA-ES also converges reliably, but computationally heavy parameters make it unsuitable for high-dimensional problems. The preferred solution depends on how all objective functions are weighed against each other, and the results show that the values of the weights will change depending on the specific problem setup. In the end, the identification of the preferred solution from the Pareto front depends on subjective decisions made by human decision makers.
| Identifer | oai:union.ndltd.org:UPSALLA1/oai:DiVA.org:uu-530444 |
| Date | January 2024 |
| Creators | Bergström, Kristina |
| Publisher | Uppsala universitet, Elektricitetslära |
| Source Sets | DiVA Archive at Upsalla University |
| Language | English |
| Detected Language | English |
| Type | Student thesis, info:eu-repo/semantics/bachelorThesis, text |
| Format | application/pdf |
| Rights | info:eu-repo/semantics/openAccess |
| Relation | UPTEC F, 1401-5757 ; 24020 |
Page generated in 0.0019 seconds