Wind farm layout optimization based on support vector regression guided genetic algorithm with consideration of participation among landowners

Abstract

Due to the existence of wake effect that causes the reduction of intake wind speed among wind turbines in the downwind direction, the wind farm efficiency is substantially discounted. An integral question to ask is how to find the optimal wind turbine layout given a wind farm. Inspired by the self-adjustment capability among individuals in the natural evolution process, a new algorithm called support vector regression guided genetic algorithm is proposed to solve the wind warm layout optimization problem which integrates the capability in each individual to adjust itself for a better “fitness” with guiding information sampled from a response surface approximated by support vector regression. It is also interesting to use the new proposed algorithm to evaluate the impact of the constraints imposed by landowners’ willingness whether to rent their land to the wind farm company. Extensive numerical experiments under different settings of wind distribution and wind farms with unusable cells are conducted to validate the proposed algorithm, shedding insights on the impact of landowners’ participation on the overall efficiency. The experiment showcases that the proposed algorithm outperforms two baseline algorithms under different conditions with improved efficiency. The proposed framework with consideration of landowners’ participation decision provide insights for wind farm planner on the different values of farm lands from the landowners.