In this chapter we introduce the multi-objective optimisation task and highlight its distinctions from the single objective optimisation task. We employ a running example of a wind farm layout optimisation problem, and illustrate how the search landscape changes depending on the objective function used, and when shifting from a single to multiple objectives. We also outline some of the different forms of uncertainty which an optimisation task may commonly experience—particularly in the context of windfarm optimisation. The chapter then continues by describing how such uncertainty can affect the performance of standard multi-objective optimisation approaches. We describe mitigation approaches which may be employed, casting the task as a robust optimisation, with different wind profile scenarios, and illustrate the use of a simple multi-objective optimiser developed for this formulation. This chapter sets the scene for later applied chapters in this volume.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Multi-objective Optimisation with Uncertainty: Some Considerations for Wind Farm Optimisation

  • Jonathan Fieldsend

摘要

In this chapter we introduce the multi-objective optimisation task and highlight its distinctions from the single objective optimisation task. We employ a running example of a wind farm layout optimisation problem, and illustrate how the search landscape changes depending on the objective function used, and when shifting from a single to multiple objectives. We also outline some of the different forms of uncertainty which an optimisation task may commonly experience—particularly in the context of windfarm optimisation. The chapter then continues by describing how such uncertainty can affect the performance of standard multi-objective optimisation approaches. We describe mitigation approaches which may be employed, casting the task as a robust optimisation, with different wind profile scenarios, and illustrate the use of a simple multi-objective optimiser developed for this formulation. This chapter sets the scene for later applied chapters in this volume.