Intelligent Learning Based Active Power Regulation of Wind Turbines Considering Fatigue Reduction

The variability and uncertainty of wind power generation pose severe challenges in balancing supply and demand in real time. To overcome this challenge, one promising solution is to exploit the self-regulation capabilities of wind turbines (WTs) and operate them as “semi-dispatchable” resources. In this article, the rotor speed control and pitch angle control are utilized simultaneously to regulate wind power output. It should be noted that changing the operating point of a WT affects the forces and moments it experienced. To increase the life span of WTs and decrease maintenance costs, an optimization problem that aims to track the dispatch command while minimize fatigue loads is formulated. Due to the nonlinearity and nonconvexity of the WT model and load estimation model, the resulting optimization problem cannot be solved online. To address this issue, an intelligent learning based method is designed. Extensive case studies are conducted, and simulation results demonstrate that the proposed method can effectively exploit the regulation capability of WTs to trace the given dispatch command while also reducing the shaft torque and thrust-induced loads.