The present invention discloses a pilot node selection method considering wind power fluctuation probability characteristics. The method comprises the following steps: counting the probability distribution, in a peak, a waist, and a valley load period, of an injection power of a wind farm in a control area; acquiring a probability distribution density function of the injection power by performingfunction fitting; adding a random injection power of the wind farm on the basis of operation modes of a peak, a waist, and a valley load to form various random operation states of a system, wherein the occurrence probabilities of the operation states are determined by the probability distribution characteristics of the injection power of each wind farm; acquiring sensitivity matrices of each operation state; by imposing random disturbance, choosing a node that, when eliminating voltage deviation of the node, the expected values of the voltage deviations of the rest load nodes under various operation states are minimal, as a pilot node. Simulation results show that, the method can effectively reflect the influence of the wind power probability characteristics on the voltage of the pilot node, and thus the method can provide guidance for the voltage control of the pilot node under the condition of large-scale wind power-grid integration.