Early fault feature of rotating machinery is weak and interfered by noise, it is difficult for accurate early fault detection. Wavelet scalogram is affected by the noise, which has low frequency resolution for the high frequency part. Reassigned wavelet scalogram is applied to improve time-frequency concentration of the scalogram. Therefore, an weak fault feature extraction method is put forward based time frequency ridge by using synchronous averaging and wavelet ridge. Firstly, multi-cycle signal is processed by continuous wavelet transform and wavelet scalogram is reassigned. Then, the scalogram is synchronous averaged. After the wavelet ridge of scalogram is extracted, instantaneous amplitude curve of signal can be calculated, where frequency spectrum analysis will be used to extract fault characteristics. Both simulations and experiments investigation have been used to verify the effectiveness of this method. It can be concluded that this method will contribute to early fault diagnosis of rotating machinery.