Date‐Driven Active Power Dispatch Method for Wind Farm Clusters Considering Health Estimation

ABSTRACT

With the rapid development of wind power technology, the expansion of wind farm scales has made it challenging to ensure safe, stable and economically efficient turbine operation. Modern wind turbines can rapidly regulate their active power output, which has driven research interest in active power control optimisation within wind farms. However, existing methods often neglect the actual health conditions of individual turbines, resulting in power allocation strategies that fail to adequately reflect each unit's operational state. This paper proposes a health‐aware active power dispatch strategy to optimise power distribution across wind farms. The approach first employs gated recurrent unit networks‐based condition monitoring approach using data from supervisory control and data acquisition system, with anomaly detection using Mahalanobis distance (MD) monitoring. Subsequently, a fuzzy evaluation algorithm quantifies the overall health status of each turbine. Finally, the active power dispatch is dynamically optimised in real‐time through a sliding time window approach, incorporating continuous health assessments of individual wind turbines. Experimental results demonstrate that compared to conventional methods, the proposed strategy achieves more sensitive fault detection and significantly reduces operational temperatures, thereby extending turbine lifespan and improving economic performance.