A new on-line approach based on rotor instantaneous power spectrum is proposed to detect doubly-fed induction generator stator winding fault. Theoretical analysis shows that

compared with the common current or voltage spectrum analysis methods

the instantaneous power spectrum has the advantage that external sensors and hardware devices are not needed. Moreover

the spectrum has good interference immunity. Once some slip tracking or calculation errors occur

the characteristic frequency will not change because the characteristic frequency is not affected by slip. A doubly-fed induction generator model is established using PSCAD software. Through normal and stator faults conditions simulations and analysis

it is concluded that the rotor instantaneous power spectrum is more sensitive than the current or voltage spectrum on dealing with the stator winding faults conditions. Finally

the experimental platform of stator faults diagnosis is built in laboratory which further proves that it is feasible to use the rotor instantaneous power spectrum to diagnose the stator winding faults.