In this paper, a time-domain analysis method based on multivariate statistic is presented for wind power generation fault diagnosis. Generally, the sound and vibration signals obtained from wind power generation are time-variant since they are strongly related to the rotational speed which is not constant even in the macro steady state. Since the mostly used signal processing method, the Fourier analysis, is only suitable for stationary signals, the development of the joint time-frequency analysis is demanded. Here, Q statistic (also referred as squared prediction error, SPE) is introduced, it is used to monitor the vibration signals and three-phase currents. The control limit of the Q statistics is calculated to decide the state of the rotating machine, and the contribution plot of SPE is used to find the fault source. The method can efficiently detect faint change and the validity of the method is proved by experiments.