An improved support vector clustering approach to dynamic aggregation of large wind farms

In this paper, we propose an improved support vector clustering (SVC) algorithm to cluster wind turbines (WTs) in a wind farm (WF). A boundary points (BPs) detecting method based on the grid theory and the connected subdomain (CS) are proposed. Thus the efficiency of SVC is enhanced while maintaining the accuracy of the algorithm. As for the multi-wind condition equivalent of WF, a method to determine the number and capacity of each of the aggregated wind turbines (AWTs) based on historical wind data is proposed. Only wind speed (WS) and wind direction (WD) of the WF are needed to calculate the WSs of each AWT. Results demonstrate that the algorithm proposed in this paper can cluster WTs quickly and accurately. And the dynamic aggregated models of the WFs are suitable for both the single-wind condition and multi-wind condition simulations, with high accuracy being obtained.