Existing methods for identification of voltage sag, cannot make full use of the voltage sampling value information and have poor adaptability to special recording conditions. Thus a homologous identification method of voltage sag events based on perceptual Hash sequence similarity is proposed. First, this paper proposes a method for extracting sag data segments based on mutation point detection and recording situation recognition. This transforms the recognition of sag events in recording files into the recognition of complete and incomplete sag data segments respectively. Secondly, it uses the Gramian angle field (GAF) to convert the extracted sag data segment from one-dimensional time series to two-dimensional images, and applies the perceptual Hash algorithm to convert it into a Hash sequence. Then, by describing the similarity through Euclidean distance, it completes the homology identification of voltage sag according to the selected threshold. Finally, it applies the simulation data generated by the IEEE30 node system to the measured data in Beijing and Fujian. This shows that the method has advantages of high recognition accuracy and good adaptability. This work is supported by the Science and Technology Guidance Project of Fujian Province (No. 2020H0009).