Multi-period reference power dispatch in offshore wind farms considering fatigue damage values of main shafts: A greedy algorithm approach

Abstract

As offshore wind turbines operate over time, they encounter cumulative fatigue damage, particularly in main shafts. This damage diminishes operational efficiency, reduces equipment lifespan, and adversely affects wind farm economic performance. To mitigate main shaft fatigue and satisfy grid power demands under variable wind conditions, this paper proposes a second-level scheduling-based reference power dispatch optimization framework with real-time fatigue awareness. First, a multiple regression model predicts main shaft torque; these predictions feed a real-time module integrating static and dynamic three-point rain-flow counting methods to assess fatigue damage. An optimization model is formulated and a greedy allocation algorithm designed to determine each turbine’s reference power dispatch scheme. Experimental results indicate the regression model achieves a 1.24 % average deviation between predicted and actual values, and cumulative static fatigue damage differs by 0.59 %. In a simulation of 100 turbines over 100 seconds, fatigue damage is reduced by 93.14 % compared to traditional average-dispatch methods, with an average computation time of 292.36 ms, demonstrating the framework’s feasibility and efficiency while satisfying real-time second-level scheduling requirements. This study provides new insights and practical solutions for effectively reducing fatigue damage in main shafts, lowering operation and maintenance costs and can extend to other components of wind turbine.