In this paper, a fault diagnosis method for wind power system converter realized by K-Nearest Neighbor algorithm (KNN) is proposed. Firstly, the mathematical model of the WPS of the Doubly-Fed Induction Generator (DFIG) with dual PWM converter is established. Then based on this model, the KNN algorithm is used for fault diagnosis. The eigenvalues used to train the fault diagnosis model are the three-phase current and capacitor voltage values under the six fault states. Finally, through the simulation verification, the results indicate that this fault diagnosis method can realize the reliable fault diagnosis of the wind power system converter described in this paper.