Planetary gear box is the key transmission equipment of wind turbine, internal structure is complex, the problems of sensor installation and signal transmission are difficult to be solved by means of vibration diagnosis. For this reason, this paper presents a method of nondestructive fault diagnosis based on stator current analysis of wind turbine generator (MCSA), and the fault features are separated and extracted by empirical mode decomposition (EMD). A simulation experimental platform of wind turbine is built based on the analysis of the structure and typical vibration characteristics of the planetary gear. The stator current signal is decomposed by EMD in the normal and simulated fault state of the planetary gear, and spectrum and instantaneous frequency analysis are performed by extracting effective IMF components. Comparative test shows, the method of MCSA can reflect the main characteristics of complex planetary gearbox faults, and the method of EMD can also be used to separate some characteristic frequency components effectively. Therefore, the combination of these two methods is helpful to widen the application scope of MCSA.