The invention relates to the technical field of intelligent power grids, and discloses a multi-feature fusion parameter power grid fault diagnosis method based on a wolf pack algorithm. The multi-feature fusion parameter power grid fault diagnosis method comprises the steps of: S1, calculating feature parameters Yi of a time domain and a frequency domain of acquired power grid data; S2, solving atarget moderate function P by adopting the wolf pack algorithm; S3, solving a fusion parameter Q of feature parameters, and solving an optimal fusion feature parameter weight coefficient combination omega; S4, and judging whether a fault exists or not. Compared with the prior art, a plurality of proper characteristic parameters of the data are fused, a plurality of feature values of mutation amount of sudden changes of a power grid can be precisely captured, deep analysis is performed on data of each point, the features of the data are analyzed more precisely, and the monitoring and diagnosisof an operation state of a power distribution network are realized.