The invention discloses a computing method of the longitudinal moment probability distribution of the power output of a wind power station, which provides a longitudinal moment analysis method, namely a method for counting wind power output at the same specific moment of every day of 365 days or more and analyzing the output fluctuation rule at the moment. Based on actually measured data, an active output probability counting result at each moment is obtained by the longitudinal moment analysis method, and is expressed by a piecewise function through function fitting. In order to verify the regularity and the applicability of the function, the data of other years is verified so that the probability distribution piecewise function at each moment has the regularity, the invariance and the applicability based on the result and can be served as the inherent attribute of the wind power output at each moment to be applied to aspects such as the reliability analysis of wind power prediction, and the planning and modifying of a scheduling plan.