Accurate interval prediction is helpful for risk analysis

enabling more reasonable decisions in power grid dispatch. Aiming at the shortcomings of conformalized quantile regression algorithm

this paper proposes a short-term wind power interval prediction method based on weighted gated recurrent conformalized quantile regression. Firstly

the quantile gated recurrent unit is used to fit the initial prediction interval in the training stage. The non-conformity and their confidence quantiles are then calculated in the calibration stage according to the non-conformity function. Then

in the test stage

the Jensen-Shannon divergency between the test set and the calibration set is calculated and weighted to the confidence quantile as the distribution weight. Therefore

a weighted confidence quantile is formed instead of the confidence quantile directly obtained from the calibration set. The final prediction interval is the algebraic sum of the initial prediction interval and the weighted confidence quantile. Finally

taking the operation data of wind farm in MengXi

China as an example

the reliability

validity and efficiency of proposed method is examined.