An advanced fatigue load model and comparison of newly developed algorithms for wind farm layout optimization

Wind energy is typically captured through wind farms. In the past, most studies on layout optimization focused on enhancing the power output, while some considered the fatigue with a simplified model. This paper develops an advanced fatigue model based on experience of individual turbines to form an innovative layout optimization framework, with maximizing the power generation and mitigating the fatigue. First, a large fatigue load database is constructed by considering a variety of inflow, wake flow, and operating conditions. Second, the database is learned using an extreme gradient-boosting tree to obtain an advanced fatigue calculation model. Finally, by integrating the fatigue load model with the Random Search algorithm (RS), a new wind farm layout optimization framework is established. To show the superiority of RS, several newly developed algorithms were compared. A validation is carried out against field data of the Horns Rev I wind farm and the discrepancy in power is 2.66 %. The new framework is applied to optimize a wind farm. The results showed that in balancing the fatigue of the wind turbines, annual energy production (AEP) still gains 0.39 %–3.89 % with a fatigue tolerance of 1–1.05 on its fatigue in normal wind conditions, respectively.