Multi-Objective Random Search Algorithm for Simultaneously Optimizing Wind Farm Layout and Number of Turbines

The Science of Making Torque from Wind

Feng, Ju^1 ; Shen, Wen Zhong^1 ; Xu, Chang^2

Department of Wind Energy, Technical University of Denmark, Lygnby

DK-2800, Denmark^1

College of Energy and Electrical Engineering, Hohai University, Nanjing

211100, China^2

关键词: Continuous variables;    Electrical cables;    Minimal spanning tree;    Multi-objective genetic algorithm;    Random search algorithm;    Wind distribution;    Wind farm developers;    Wind farm layout optimizations;   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/753/3/032011/pdf
DOI  :  10.1088/1742-6596/753/3/032011

来源: IOP

【 摘 要 】

A new algorithm for multi-objective wind farm layout optimization is presented. It formulates the wind turbine locations as continuous variables and is capable of optimizing the number of turbines and their locations in the wind farm simultaneously. Two objectives are considered. One is to maximize the total power production, which is calculated by considering the wake effects using the Jensen wake model combined with the local wind distribution. The other is to minimize the total electrical cable length. This length is assumed to be the total length of the minimal spanning tree that connects all turbines and is calculated by using Prim's algorithm. Constraints on wind farm boundary and wind turbine proximity are also considered. An ideal test case shows the proposed algorithm largely outperforms a famous multi-objective genetic algorithm (NSGA-II). In the real test case based on the Horn Rev 1 wind farm, the algorithm also obtains useful Pareto frontiers and provides a wide range of Pareto optimal layouts with different numbers of turbines for a real-life wind farm developer.