Multi-objective Optimisation of Floating Offshore Wind Farms based on a Real-World Case Study

In this paper, we used real data to predict the position of the floating wind turbines due to three external forces: wind, wave and current. We analysed data provided by energy company Equinor and applied two machine-learning techniques: Multilayer Perceptron and Random Forest. After demonstrating that machine learning models failed, we used a simple linear regression model and optimisation approach to solve the multi-objective optimisation problem. We minimised the perimeter of the wind farm and maximised its power output by applying the multi-objective optimisation algorithm NSGA-II. We also investigated how changing the length of mooring lines affected the optimisation. We used hypervolume to measure the algorithm’s performance. We have shown that the results are very similar for fixed and floating wind farms. However, we have found that in complex wind conditions, i.e., when the wind does not blow only from one direction, for many wind turbines, the floating wind farms were a better solution than the fixed ones.