Scholarlink

<p>In wind farms, the wake effect from upstream wind turbines (WTs) reduces the efficiency of downstream counterparts and overall power generation. Yaw control of upstream WTs can mitigate wake effects and maximize power output, but it must also consider structural impacts to prevent excessive fatigue loads. This paper presents an optimization framework for managing both power and fatigue loads in wind farms using machine learning and multi-objective optimization algorithms. The framework aims to maximize power generation while minimizing fatigue loads. Simulations under complex inflow conditions and machine learning methods are used to rapidly predict power and fatigue loads. The non-dominated sorting genetic algorithm III then optimizes the yaw angles of WTs to achieve optimal power output with minimal fatigue loads, enhancing wind farm efficiency over its lifetime. Results show that the proposed strategy increases power by 8.4% compared to no yaw control and reduces fatigue loads by 12.5% and 4.8% under different objectives, with only a 1.5% reduction in power compared to power-maximizing yaw control. This framework can be extended to other multi-objective optimization problems in wind farms.</p>

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

Optimization of wind farm power and fatigue based on machine learning

  • 期刊论文

  • 发表日期：

    2025年4月30日

查看全文
  • Yuntao Gao,
  • Lei Wang,
  • Zhiqiang Xin,
  • Quanlin Ding,
  • Hao Wang

摘要

In wind farms, the wake effect from upstream wind turbines (WTs) reduces the efficiency of downstream counterparts and overall power generation. Yaw control of upstream WTs can mitigate wake effects and maximize power output, but it must also consider structural impacts to prevent excessive fatigue loads. This paper presents an optimization framework for managing both power and fatigue loads in wind farms using machine learning and multi-objective optimization algorithms. The framework aims to maximize power generation while minimizing fatigue loads. Simulations under complex inflow conditions and machine learning methods are used to rapidly predict power and fatigue loads. The non-dominated sorting genetic algorithm III then optimizes the yaw angles of WTs to achieve optimal power output with minimal fatigue loads, enhancing wind farm efficiency over its lifetime. Results show that the proposed strategy increases power by 8.4% compared to no yaw control and reduces fatigue loads by 12.5% and 4.8% under different objectives, with only a 1.5% reduction in power compared to power-maximizing yaw control. This framework can be extended to other multi-objective optimization problems in wind farms.