The invention discloses a quantile regression-based wind power fluctuation interval analysis method. The method comprises the steps of obtaining output power data of a wind power plant firstly; determining the quantiles of the output power data of the wind power plant; establishing a regression function and a regression model for each quantile by adopting a support vector machine; next, solving the regression model of each quantile by adopting a prime-dual interior point method, and calculating the quantile in the next moment; S6: obtaining the actually measured value of the wind power data atthe next moment; and finally, returning to the repeated cycle to obtain the output power data fluctuation interval of the wind power plant. The quantile regression-based wind power fluctuation interval analysis method provided by the invention can realize the adaptive selection of the regression functions through the support vector machine without making any assumed conditions on the random disturbance terms, so as to determine the quantile regression models; next, the models are solved by the prime-dual interior point method with infeasible initial points, so that wind power fluctuation interval analysis for the further moments is realized; and by virtue of the method, the complete wind power fluctuation interval analysis result can be obtained, and the newest change condition also can be reflected in real time.