The invention discloses a virtual wind measurement tower technology-based ultra-short-term power prediction method for a wind power plant. The method comprises the steps of 1, collecting basic data, historical data and real-time data of n fans of the wind power plant; 2, performing analog computation on a wind speed and a wind direction at a hub height of the whole wind power plant at a (zeta)th moment; 3, performing analog computation on a temperature at the hub height of the whole wind power plant at the (zeta)th moment; 4, computing air density, humidity and air pressure of the wind power plant at the (zeta)th moment; 5, obtaining virtual wind measurement tower data of the wind power plant at the (zeta)th moment; 6, obtaining a historical virtual wind measurement tower data set of the wind power plant according to the steps 1-5; 7, building a support vector machine model for ultra-short-term power prediction of the wind power plant and training the model by using the historical virtual wind measurement tower data set; and 8, substituting the virtual wind measurement tower data of the wind power plant at the (zeta)th moment into the trained model, thereby obtaining an ultra-short-term power prediction value data set of the wind power plant at the (zeta)th moment. The problems of relatively low ultra-short-term power prediction precision and the like of the wind power plant in the prior art are solved.