In order to accurately identify all kinds of sag sources in power grid, a method based on improved grey correlation analysis is presented for identification of voltage sags source. Firstly, the generation mechanism of different voltage sag is analyzed, and the waveform characteristics of various sag sources in power grid are analyzed by using the sag segmentation method. Secondly, considering the shortcomings of the traditional grey correlation analysis model, the entropy weight method is used to improve it. Then, extract the time domain characteristics of voltage sag waveform to form the standard reference sequence of six types of sag sources and the comparison sequence corresponding to the sag sources to be identified. The improved grey correlation analysis model is used to calculate the correlation degree between the reference sequence and comparison sequence to achieve accurate identification of sag sources. Finally, the proposed method is validated by PSCAD/EMTDC simulation data and measured data. And compared with other existing methods, the proposed method can not only accurately identify all kinds of sag source with fewer samples, but can also determine the specific fault types while sags are caused by short circuit faults. It has great engineering application prospects.